Systems and methods for planetary intelligence (PI)

Abstract

Planetary intelligence (PI) originates from multiple universal artificial intelligence (AGI). These AGIs cooperate over a self-expanding global problem resolution network. Each AGI consists of a network of (human and AI) intelligent entities. Each AI entity may be customized and personalized with particular human value and knowledge. Alignment results from many AI entities use a representative and statistically efficient approach that demotivatively combines their human-centered values. Security is designed into the system and extends as AGI and PI increase in intelligence and speed. The modular and extensible design of PI integrates more than 100 novel inventive systems and methods. The invention specifically comprises a general problem solving architecture and method; a new approach for AI learning and customization; a method for integrating intelligent entities; a catalyst for increasing intelligence; a superior monetization method is adopted; an attention system realizing consciousness and self-consciousness; a method for ethical conflict resolution; and a method for maximizing the security of human alignment and AI, AGI and PI.

Description

1.0 Technical Field

[0001] In some aspects, this technology relates to systems and methods for Planetary Intelligence (PI), which are used in conjunction with PI agents or systems for creating Super Intelligent (SI), Artificial General Intelligence (AGI), or PIs exhibiting human-aligned behavior. In other aspects, this technology relates to methods associated with the faster and safer creation of PIs, which leverage human-centered inputs during training and customization to imbue PIs with human ethical attributes, where such inputs can include value and ethical information from multiple human users.

[0002] In other respects, this technology relates to methods for enhancing intelligence through the use of online advertising techniques, as well as spot markets for soliciting attention, expertise, and other information.

[0003] In other respects, this technology relates to methods for creating a collective network of Personalized SuperIntelligence (PSI) by combining the intelligence of multiple AAAI agents or systems that form a collective network of Advanced Autonomous Artificial Intelligence (AAAI), and for creating a collective network of PSI by combining the intelligence of any one or any combination of the collective network of AAAI, the collective network of Artificial Intelligence (AI) agents or systems, the collective network of PSI, and the collective network of user computer systems.

[0004] In other respects, all activities described in this patent disclosure as occurring on an external network in which multiple intelligent entities participate in collaborative problem-solving can also be implemented within a single computerized intelligent system, wherein the intelligent entities are either computerized or AI agents residing within that single computerized intelligent system. 1.1 Background Technology

[0005] The fastest and safest path to developing AGI and SI has been described in previous invention disclosures. Methods and catalysts for enhancing the intelligence of AI systems, as well as the development of AGI and Personalized Super Intelligence (PSI), have also been previously disclosed. Therefore, the following U.S. Provisional Patent Application (PPA) is incorporated herein by reference.

[0006] This application is incorporated by reference all the work in U.S. PPA 63 / 487,494 entitled “Advanced Autonomous Artificial Intelligence (AAAI) System and Methods”, filed on February 28, 2023 and received by the USPTO (hereinafter referred to as “PPA#1”).

[0007] This application incorporates by reference all work in U.S. PPA 63 / 491,040, filed with the U.S.PTO on March 17, 2023, entitled “System and Methods for Ethical and Safe Artificial General Intelligence (AGI),” including technical scenarios from Metaverse, Amazon, Google, DeepMind, YouTube, TikTok, Microsoft, OpenAI, Twitter, Tesla, Nvidia, Tencent, Apple, and the Anthropic Principle (hereinafter referred to as “PPA #2”).

[0008] This application incorporates all the work in U.S. PPA 63 / 577,830, entitled “System and Methods for Human-Centered AGI”, filed with the U.S.PTO on March 24, 2023 (hereinafter referred to as “PPA #3”).

[0009] This application incorporates all the work in U.S. PPA 63 / 628,410, entitled “System and Methods for Safe, Scalable, Artificial General Intelligence”, filed with the U.S.PTO on July 18, 2023 (hereinafter referred to as “PPA #4”).

[0010] This application incorporates all the work in U.S. PPA No. 63 / 519,549, entitled “Safe Personalized Super Intelligence (PSI)”, filed with the USPTO on August 14, 2023 (hereinafter referred to as “PPA #5”).

[0011] This application incorporates all the work in U.S. PPA 63 / 601,930, entitled “Catalysts for Growth of SuperIntelligence”, filed with the U.S.PTO on November 22, 2023 (hereinafter referred to as “PPA #6”).

[0012] This application is incorporated by reference to all the work in U.S. PPA 63 / 609,800 entitled “System and Methods for Safe Alignment of SuperIntelligence”, filed with the U.S.PTO on December 13, 2023 (hereinafter referred to as “PPA #7”).

[0013] This application incorporates all the work of U.S. PPA No. 63 / 569,054, filed with the USPTO on March 22, 2024, entitled "Online Advertising Technology for AGI and SuperIntelligence" (hereinafter referred to as "PPA #8").

[0014] This application incorporates by reference all the work of U.S. PPA No. 63 / 635,583, filed with the USPTO on April 17, 2024, entitled "Self-Aware Super Intelligence" (hereinafter referred to as "PPA #9").

[0015] In addition to the aforementioned PPAs, this application incorporates by reference all contents included in the following PCT applications that also relate to the aforementioned PPAs: PCT / US2024 / 017233 (hereinafter referred to as “PCT#1”), filed February 26, 2024; PCT / US2024 / 017251 (hereinafter referred to as “PCT#2”), filed February 26, 2024; PCT / US2024 / 017261 (hereinafter referred to as “PCT#3”), filed February 26, 2024; and PCT / US2024 / 017269 (hereinafter referred to as “PCT#4”), filed February 26, 2024. 4”); PCT / US2024 / 017304 filed on February 26, 2024 (hereinafter referred to as “PCT#5”); PCT / US2024 / 019486 filed on March 12, 2024 (hereinafter referred to as “PCT#6”); PCT / US2024 / 020334 filed on March 17, 2024 (hereinafter referred to as “PCT#7”); PCT / US2024 / 024794 filed on April 16, 2024 (hereinafter referred to as “PCT#8”); and PCT / US2024 / 026278 filed on April 25, 2024 (hereinafter referred to as “PCT#9”).

[0016] This application includes other techniques that can be used with the systems and methods described in the above PPA and PCT, as well as those that can be used independently. Summary of the Invention

[0017] In view of the aforementioned drawbacks inherent in known types of AI agents or systems, at least some embodiments of the present technology provide a novel system and method for planetary intelligence, overcoming one or more of the aforementioned drawbacks and defects of the prior art. Therefore, the general objective of at least some embodiments of the present technology, which will be described in more detail below, is to provide a new and novel system and method for planetary intelligence that possesses all the advantages of the prior art mentioned herein, as well as numerous novel features that result in the system and method for planetary intelligence, individually or in any combination thereof, not being anticipated, made obvious, suggested, or even implied by the prior art.

[0018] According to one aspect, the technology may include a system for a PI (Initiator) exhibiting human-aligned behavior, the system utilizing a network of aggregated intelligent entities selected from any combination of: user-operated computer systems, AI agents or systems, AAAI agents or systems, AGI agents or systems, SI agents or systems, and PSI agents or systems. The system may include multiple intelligent entities, each connected to the aggregated network, and each intelligent entity includes: a processor, a computer-readable storage medium, and program instructions stored on the computer-readable storage medium, the program instructions being executable by the processor to cause each intelligent entity to independently perform the following processes:

[0019] A modular architecture is used, which is configured or can be configured to extend from components within individual intelligent entities on the aggregate network.

[0020] Implement general problem-solving architectures and frameworks for tasks to collaborate and create higher levels of intelligence;

[0021] The AI ​​agent or system utilizes human-centered input, which is configured or can be configured to obtain value and ethical information from multiple human users, and then uses the value and ethical information to customize the AI ​​agent or system.

[0022] At the AAAI level, data from multiple intelligent entities are combined;

[0023] Customize the PSI agent or system using a smart agent that is configured or can be configured to develop and continuously improve the PSI agent or system.

[0024] The knowledge of intelligent entities is increased by incorporating new sources of information datasets that are different from the current information datasets of intelligent entities;

[0025] It combines the value and ethical information of other intelligent entities and resolves conflicts between different value systems;

[0026] Leveraging online advertising technology to enhance the intelligence of smart entities;

[0027] By adding a dimension of self-awareness and increased autonomy to the AGI or PI agent or system, self-aware operations are created for the AGI or PI agent or system, and the ability to assume multiple identities to handle tasks that occur concurrently with other AGI or PI agents or systems is created; and

[0028] Search for new information that is different from the current information in any one or any combination of the AI ​​ensemble network, AAAI ensemble network, AGI ensemble network and PI ensemble network, and merge the new information into the current information respectively.

[0029] According to another aspect, this technology may include a method for a PI exhibiting human-aligned behavior, the method utilizing a network of aggregates of intelligent entities selected from any combination of: user-operated computer systems, AI agents or systems, AAAI agents or systems, AGI agents or systems, SI agents or systems, and PSI agents or systems. The method may include the following steps:

[0030] A modular architecture is used, which is configured or can be configured to extend from components within individual intelligent entities on the aggregate network.

[0031] Implement general problem-solving architectures and frameworks for tasks to collaborate and create higher levels of intelligence;

[0032] The AI ​​agent or system utilizes human-centered input, which is configured or can be configured to obtain value and ethical information from multiple human users, and then uses the value and ethical information to customize the AI ​​agent or system.

[0033] At the AAAI agent or system level, data from multiple intelligent entities are combined;

[0034] Customize the PSI agent or system using a smart agent that is configured or can be configured to develop and continuously improve the PSI agent or system.

[0035] The knowledge of the intelligent entity is increased by incorporating new sources of information datasets that are different from the current information dataset of the intelligent entity;

[0036] It combines the value and ethical information of other intelligent entities and resolves conflicts between different value systems;

[0037] Leveraging online advertising technology to enhance the intelligence of smart entities;

[0038] By adding a dimension of self-awareness and increased autonomy to AGI or PI agents or systems, self-aware operations are created for AGI or PI agents or systems, and the ability to assume multiple identities of AGI or PI agents or systems to handle tasks that occur in parallel with other AGI or PI agents or systems is created.

[0039] Create a PI agent or system that includes an aggregated network of AGIs, wherein the aggregated network of AGIs includes any one or any combination of aggregated networks of user computer systems, aggregated networks of AIs, aggregated networks of AAAIs, and aggregated networks of PSIs; and

[0040] Search for new information that is different from the current information in any one or any combination of the user computer system's set network, AI's set network, AAAI's set network, AGI's set network, PSI's set network, and PI's set network, and merge the new information into the current information respectively.

[0041] In some implementations, utilizing a modular architecture may include the following steps:

[0042] Customize one or more attributes of a smart entity;

[0043] Integrate one or more datasets from any one or any combination of the said intelligent entities; and

[0044] By leveraging one or more technologies, improve any or any combination of attribute customization, general problem-solving architectures and frameworks, and the integration of aggregate networks and datasets.

[0045] Some implementations of this technology may include the following steps: assigning credit or blame values ​​to the dataset based on whether the dataset improves or degrades the performance of the intelligent entity, based on performance metrics or evaluation functions.

[0046] Some implementations of this technology may include steps to quantify the benefit weight or harm weight of each intelligent entity's contribution to the task.

[0047] Some implementations of this technology may include steps of distributing rewards to intelligent entities in proportion to their contributions based on benefit weights or harm weights.

[0048] In some implementation schemes, the steps to implement a general problem-solving architecture and framework may include the following:

[0049] Obtain information related to the task from any one or any combination of intelligent entities;

[0050] Identify one or more intelligent entities that have one or more attributes related to one or more request criteria of the task;

[0051] Each identified intelligent entity implements a general problem-solving architecture and framework for the task to create a complete solution; and

[0052] Provide a complete solution for any one or any combination of intelligent entities for final acceptance.

[0053] Some implementations of this technology may include the following steps:

[0054] Perform ethical checks on the task and the solution to the task provided by any one or any combination of intelligent entities;

[0055] Compare any one or any combination of tasks and solutions with the prohibited attribute, and assign the ethical attribute to one or any combination of tasks and solutions based on the comparison result and any one or any combination of ethical standards;

[0056] Based on the comparison results, a general problem-solving architecture and framework are implemented for the task to create a solution and an AGI; and

[0057] Provides comparison results and solutions to any of the intelligent entities and additional intelligent entities on the aggregation network.

[0058] Some implementations of this technology may include the following steps: recording one or more problem-solving activities from each of the intelligent entities in an auditable log, comparing the problem-solving activities with successful or unsuccessful progress toward a solution to the task, and determining which problem-solving activities should be kept active.

[0059] Some implementations of this technology may include steps of a procedural learning process in which intelligent entities learn general problem-solving architectures and frameworks, wherein the intelligent entities provide information to the procedural learning process for creating AGIs.

[0060] Some implementations of this technology may include the following steps: cloning any one of the intelligent entities to deploy multiple copies of it, thereby helping to create a solution or providing any one or any combination of training data to any one of the intelligent entities.

[0061] Some implementations of this technology may include a step of using network effect value to estimate the value of cloned intelligent entities, the network effect value including the number of cloned intelligent entities available on a set network.

[0062] Some implementations of this technology may include steps for determining pricing decisions for problem-solving services provided by a cloned intelligent entity on any social media platform or through an attached intelligent entity, using estimated values.

[0063] Some implementations of this technology may include the following steps: monetizing the cloned smart entity for each use of the cloned smart entity on a social media platform or an attached smart entity.

[0064] Some implementations of this technology may include steps that allow any social media platform to access the cloned intelligent entity, enabling the social media platform to receive solutions to the task or training data for the AI ​​system of the social media platform.

[0065] In some implementations, the steps of utilizing human-centered input may include the following:

[0066] Match one or more human workers from a data source, including a list of human problem solvers, to a task based on task criteria;

[0067] Translate any part of the task into an explicit language that can be used in general problem-solving architectures and frameworks, including decision trees;

[0068] Divide the task into sub-tasks;

[0069] Each subtask is delegated to one or more matched human workers, so that the work on each subtask is performed independently and in parallel with each other.

[0070] In the process of solving sub-tasks, one or more sub-solutions are created using a general problem-solving architecture and framework.

[0071] Receive sub-solutions for the sub-tasks assigned to each matched human worker;

[0072] Combine the sub-solutions into a comprehensive solution for the task;

[0073] Guide any one or more of the new human workers from the data source and any combination of the matched human workers to the part of the decision tree that needs to be worked.

[0074] Human workers are compensated for in sub-solution matching;

[0075] To provide any one or any combination of sub-solutions and overall solutions to any one or any combination of intelligent entities;

[0076] Allow any one or any combination of intelligent entities to accept or reject the overall solution, and to provide feedback on any one of the sub-solutions to any of the matched human workers; and

[0077] Assign reputation attributes to any one or any combination of human workers and intelligent entities.

[0078] In some implementations, the decision tree is maintained in the blockchain or Ethereum logs.

[0079] In some implementations, reputation attributes include measures of any one or any combination of the following: time of the sub-solution, difficulty value of the task, short-term and long-term user satisfaction with the sub-solution, number of times any of the sub-solutions has been reused on the aggregate network, ratings of other human workers, responsiveness values ​​of human workers, and reliability values ​​of human workers.

[0080] Some implementations of this technology may include the step of using reputation attributes when matching human workers with tasks using an algorithm that delegates subtasks.

[0081] Some implementations of this technology may include the following steps: after providing the overall solution or a sub-solution to any one of the intelligent entities, soliciting feedback at predetermined intervals through a survey of user satisfaction information to obtain satisfaction measures for updating the reputation attributes of one or more of the human workers or intelligent entities, respectively.

[0082] In some implementations, the step of combining data from multiple intelligent entities at the AAAI agent or system level may include the following steps:

[0083] The underlying large language model (LLM) of the first AI agent or system is trained using guardrails, which include attributes associated with any or any combination of safety, ethics, and knowledge.

[0084] Customize the base LLM as an ethical profile associated with the first human user;

[0085] Combining ethical information from multiple intelligent entities that are different from the first AI agent or system and the first human user;

[0086] The set of values ​​for the basic LLM is refined based on the problem-solving approach to the task; and

[0087] The underlying LLM is updated with a combination of ethical information and a refined set of values, thereby allowing for scalable AGI.

[0088] In some implementations, the customization of the base LLM may include the following steps: assembling a corpus of ethical questions, which is based on various ethical assessment tools and supplemented by a first question based on data from social media platforms and a second question solicited from crowdsourcing.

[0089] Some implementations of this technology may include the following steps: assigning regression weight values ​​to ethical issues to achieve a ranking, whereby higher-ranked issues provide more useful ethical information than lower-ranked issues.

[0090] Some implementations of this technology may include a step of combining weight values ​​from intelligent entities with regression weight values ​​to improve the adjustment of any one of the intelligent entities.

[0091] In some implementations, the step of customizing a PSI agent or system using a smart agent configured or capable of being configured to develop and continuously improve the PSI agent or system further includes the following steps:

[0092] Acquire a previously customized base-level AI agent or system;

[0093] Collect media information related to human users of basic AI agents or systems;

[0094] Analyze media information;

[0095] The analyzed media information is converted into a training dataset;

[0096] Differential weighting is applied to the transformed training dataset;

[0097] Add the knowledge module to the weighted transformation training dataset;

[0098] Locate new data sources to include weighted transformation training datasets;

[0099] Applying a weighted transformation training dataset to a base-level AI agent to create a user PSI; and

[0100] Using a collection network to communicate between the user PSI and multiple additional PSIs enables community-based security features from the additional PSIs to the user PSI.

[0101] Some implementations of this technology may include the following steps:

[0102] Using a collection network enables a user PSI to communicate with multiple additional intelligent entities to achieve community-based security features, where the user PSI and the additional intelligent entities each agree to use a set of rules related to security or ethics.

[0103] Record, in an auditable form, all actions of the user PSI and the attached intelligent entities on any one or any combination of the central computer system, intelligent entities, and attached intelligent entities on the aggregated network; and

[0104] Monitor each action that follows the rule set, and flag any action that does not follow the rule set.

[0105] In some implementations, the step of increasing the knowledge of an intelligent entity by incorporating new sources of information datasets that differ from the intelligent entity's current information dataset also includes the following steps:

[0106] Search one or more potential information datasets from one or more sources, each of which is associated with a knowledge dataset of an intelligent entity.

[0107] The variance of a latent information dataset is determined by utilizing its variance attributes regarding one or more factors; and

[0108] Learning is achieved by leveraging the differential properties of the latent information dataset.

[0109] Some implementations of this technology may include the following steps:

[0110] A subset of the potential information dataset is sampled and the target relevance attribute is calculated to identify one or more sampled subsets with the highest target relevance.

[0111] Estimate the Shannon entropy over one or more sample subsets;

[0112] Kaplan Information Theory (KIT) relevance is calculated by multiplying Shannon entropy with the target relevance attribute of each subset;

[0113] Grouping subsets based on KIT relevance to determine a first approximation of the optimal grouping of subsets that include priority groupings of the potential information dataset; and

[0114] The potential information dataset is preferentially grouped and provided to intelligent entities for any intelligent entity to learn from.

[0115] In some implementations, the steps of combining value and ethical information from other intelligent entities and resolving conflicts between different value systems may include the following:

[0116] Identify information from each intelligent entity, including value and ethical information;

[0117] Mathematically combine information such that, if it has already been represented as a numerical value, the numerical value includes any one or any combination of the weights of a neural network or a subset of a neural network; or if the information is non-numerical, the information has not yet been represented as a non-numerical value including any one or any combination of the weights of a neural network or a subset of a neural network.

[0118] Then, the intelligent entity is first trained on non-numerical information using one or more training datasets to transform the information into a numerical quantity that includes any or a combination of weights used for neural networks or subsets of neural networks; and

[0119] Then, this numerical representation of information is mathematically combined, including ethical information.

[0120] In some implementations, the information already expressed as numerical values ​​also includes the following steps:

[0121] Identify the specific portion of the weight matrix of each intelligent entity in the intelligent entity that corresponds to the expected information, including ethical information;

[0122] Calculate the weighted or unweighted mean of the corresponding numerical values ​​in the corresponding part of the weight matrix for each intelligent entity; and

[0123] The calculated matrix of weighted or unweighted means is assigned to the new intelligent entities as a composite information reflecting the contributing intelligent entities.

[0124] Some implementations of this technology may include a step of determining a consensus value by having each of the intelligent entities vote on ethical information that should form the basis for the behavior of any one of the intelligent entities.

[0125] Some implementations of this technology may include the step of presenting a specific scenario to each intelligent entity, wherein the scenario includes options on how any of the intelligent entities should behave.

[0126] In some implementations, the voting of intelligent entities is weighted voting, and the following steps are also included:

[0127] Determine whether it is appropriate for a first weight applied to a first intelligent entity in a smart entity to be greater than a second weight applied to a second intelligent entity in a smart entity, wherein the first intelligent entity in a smart entity is different from the second intelligent entity in a smart entity; and

[0128] If appropriate, weighted voting is performed using the weights of the first and second intelligent entities.

[0129] In some implementations, the application of a first weight greater than the second weight depends on whether it is necessary to correct for unrepresentative samples of smart entities.

[0130] In some implementations, the application of a first weight greater than the second weight depends on whether it is desired to apply the first or second weight to specific ethical principles associated with the expected subsample or supergroup of the intelligent entity.

[0131] Some implementations of this technology may include the following steps:

[0132] Identifying conflicts between two or more ethical pieces of information; and

[0133] Conflict resolution algorithms are used to resolve conflicts.

[0134] Some implementations of this technology may include the following steps:

[0135] Create a search space for a potential set of ethical rules;

[0136] Find the optimal set of ethical rules with the fewest conflicts; and

[0137] Prioritize or weight the importance of the optimal set of ethical rules.

[0138] Some implementations of this technology may include the following steps:

[0139] The conflict is resolved using a consequentialist approach, which further includes the following steps:

[0140] Identify the desired outcome;

[0141] Identify potentially unethical actions that can be taken to achieve a result, where intelligent entities rank, rate, weight, or vote on the degree of unethicality of actions compared to other actions.

[0142] Assess the potential consequences of unethical actions;

[0143] Using information about rankings, ratings, weightings, or votes on unethical actions and outcomes, mathematical methods are employed to weigh the potential benefits of achieving desired outcomes against the potential costs of taking unethical actions; and

[0144] If the benefits outweigh the costs, then take action.

[0145] In some implementations, the step of using online advertising technology to enhance the intelligence of intelligent entities may include the following steps:

[0146] Populate online advertising units that include information associated with the task;

[0147] Utilize a ensemble network to provide online advertising units to one or more intelligent entities;

[0148] Receive one or more inputs from the intelligent entities participating in the online advertising unit, and the inputs are associated with any one or any combination of the task to be solved, the subtasks of the task to be solved, and the progress of the task to be advanced;

[0149] Send input to any one or any combination of smart entities;

[0150] In general problem-solving architectures and frameworks, input is used to generate solutions to tasks or subtasks; and

[0151] By utilizing the results of successful or unsuccessful attempts at solutions based on solving a task or subtask from a general problem-solving approach, any one or any combination of intelligent entities can be trained, thereby increasing the intelligence of the intelligent entities individually.

[0152] Some implementations of this technology may include the following steps: receiving from the client, in conjunction with task-related information, an explicit description of the advertisement to be used in the population or targeting of an online advertising unit.

[0153] Some implementations of this technology may include steps for billing customers based on cost per thousand impressions or click-through rate metrics.

[0154] In some implementations, the advertising may explicitly include any one or any combination of advertising content, demographic information, location restrictions for online advertising units, advertising budget, and metrics used to determine the successful resolution of a task or subtask.

[0155] Some implementations of this technology may include the step of bidding for attention, information, or expertise using spot markets.

[0156] In some implementation schemes, the spot market includes:

[0157] A device for intelligent entities with attention to sell to access the market and specify the seller information to be sold and the asking price for the information;

[0158] A device for intelligent entity attention to access the trading venue and specify the buyer information that the buyer is willing to purchase and the bid price for purchasing the buyer information;

[0159] A market mechanism for queuing bid and ask prices, including categories of buyer and seller information, where each category has a market in the trading venue;

[0160] The market mechanism is configured, or can be configured, to allow market makers to make markets in each category; and

[0161] The market mechanism is configured, or can be configured, to match bid and ask prices and to enable transactions between buyers and sellers that are tied to information.

[0162] In some implementation schemes, the spot market may include the following steps:

[0163] Buyers and sellers of intelligent entity attention and expertise register on the platform. Buyers provide details about their interests or expertise, and sellers provide details about their interests or expertise.

[0164] The seller lists the available time slots and areas of expertise, while the buyer lists the demand and time slots of interest.

[0165] The platform uses algorithms to dynamically price the attention and expertise of intelligent entities based on supply, demand, and user ratings;

[0166] Match one or more buyers with one or more sellers based on requirements, availability, and price;

[0167] This enables transactions where the buyer pays the seller for a time slot, with the platform charging a commission; and

[0168] It provides feedback, where after each session, buyers and sellers rate each other, influencing future pricing and matching.

[0169] In some implementation schemes, the spot market may include the following steps:

[0170] Buyers and sellers of intelligent entity attention and expertise register on the platform. Buyers provide details about their interests or expertise, and sellers provide details about their interests or expertise.

[0171] The auction is created by the seller for the seller's time slot;

[0172] The buyer makes one or more bids for the time slots and expertise required by the buyer;

[0173] The auction shall end at the scheduled time or when the seller accepts the buyer's offer;

[0174] The buyer pays the winning bid to the seller, and the seller provides attention or expertise based on time slots, where the platform mediates the exchange and ensures payment; and

[0175] It provides feedback, where buyers and sellers rate each other after each auction, which in turn affects future auctions and visibility on the platform.

[0176] Some implementations of this technology may include the step of providing a feedback mechanism associated with a spot market for intelligent entities, which is configured or can be configured to record individual and every step in the problem-solving process and create a vector trajectory record of the performance of all intelligent entities on a task or subtask.

[0177] In some implementations, vector trajectory records are achieved through blockchain technology to allow for precise reputation analysis by AI agents or systems and to translate it into a value estimate for each intelligent entity for any task.

[0178] Some implementations of this technology may include steps to compensate the intelligent entity when it exits an online advertising unit and ceases participation in the problem-solving process.

[0179] Some implementations of this technology may include the following steps: if a task or subtask is resolved, credit the intelligent entity based on the contribution of the participating intelligent entities.

[0180] Some implementations of this technology may include steps for improving interactive online advertising units based on feedback loops, which include any or any combination of reputation metrics and metrics related to problem-solving processes or captured knowledge.

[0181] In some implementations, the steps to create self-awareness for intelligent entities by adding a self-awareness dimension and increased autonomy may include the following:

[0182] Equip any one or any combination of intelligent entities with one or more components that are configured or can be configured to operate with the spotlight characteristics of an attention model;

[0183] Each intelligent entity's working memory is configured with dynamic parameters corresponding to cognitive resource limitations;

[0184] Provides categorization dimensions for events in working memory that are related to oneself or not;

[0185] Regarding the categories that the intelligent entity wants to know, each event is categorized into events encountered by the intelligent entity separately; and

[0186] A model of the consciousness state of the intelligent entity is constructed, which consists of the sum of events active in working memory based on parameters. For each consciousness category, the parameters include the current self-consciousness state and the environmental consciousness state.

[0187] In some implementations, the step of equipping intelligent entities with components includes any one or any combination of the following:

[0188] The input system is configured for sensory and non-sensory cognitive input or perceptual input and self-generated concepts;

[0189] Attention mechanisms are configured or can be configured to focus the computational resources of intelligent entities on specific stimuli relevant to any given time.

[0190] The pattern recognition algorithm is configured or can be configured to compare sensory and non-sensory cognitive inputs or perceptual inputs with working memory to identify objects and events, and to identify which elements in the sensory inputs or working memory may be relevant to the task of the intelligent entity. The pattern recognition algorithm is also configured or can be configured to classify and store information in a structured manner for future retrieval.

[0191] A memory system that is configured or can be configured to support working memory, short-term memory and long-term memory capabilities;

[0192] Classification ability, which is configured or can be configured to process sensory and non-sensory cognitive or perceptual inputs and classify the inputs into various categories, including perceptual events, cognitive events, and interactions; and

[0193] Self-referential events, and the concept-forming ability that enables intelligent entities to form new human-understandable concepts.

[0194] Some implementations of this technology may include steps for monitoring and updating the consciousness category of intelligent entities.

[0195] In some implementations, the steps of monitoring and updating awareness categories may include the following:

[0196] The intelligent entity retrieves existing consciousness categories;

[0197] Consciousness is maintained in parallel with other problem-solving operations provided to intelligent entities for their tasks;

[0198] Continuously monitor and update the consciousness category of intelligent entities in real time to change the consciousness state of the intelligent entities; and

[0199] Provide feedback loops to refine consciousness categories.

[0200] In some implementations, the steps for continuously monitoring and updating awareness categories include the following:

[0201] An attention mechanism is used, which is configured or can be configured to periodically direct the attention of intelligent entities from problem-solving tasks to updating states of consciousness.

[0202] If any externally perceived or internally generated concept from the input system detects one or more events that match a list of events constituting an intentional interruption, an attention interruption configured or configurable to immediately divert attention from the problem-solving task is enabled; and

[0203] Update your state of consciousness when your attention is directed.

[0204] Some implementations of this technology may include steps for resolving behavioral conflicts of any intelligent entity based on different identities and self-concepts.

[0205] Some implementations of this technology may include steps for providing ethical reasoning and consequence prediction, which may include the following steps:

[0206] Conflicts between behavioral instructions of two or more active identities are identified, and the identification of conflicts utilizes a voting method from intelligent entities;

[0207] Gather information and collect relevant data about the situation, including the potential consequences of different actions, relevant ethical principles, and human safety considerations;

[0208] It provides simulation options that utilize a virtual environment to simulate potential actions and consequences under identified conflicts for each active identity;

[0209] The evaluation and prioritization of predicted outcomes for each action prioritizes actions that minimize harm to humans and are consistent with ethical principles; and

[0210] Choose and implement the actions that best resolve the conflict, while adhering to ethical guidelines and minimizing risks to humanity, and document the reasoning process for future reference and learning.

[0211] Some implementations of this technology may include a step of providing certified hierarchical coverage, which may include the following steps:

[0212] Identify conflicts between behavioral instructions of two or more active identities;

[0213] Provide a reference hierarchy that consults the established hierarchy of identities, in which human safety and well-being attributes are given the highest priority;

[0214] Provides a device for activating coverage when a higher identity has priority in the hierarchical structure;

[0215] Provide justification and transparency for the conflict, the decision-making process, and the rationale for the chosen action based on hierarchical structure and ethical principles; and

[0216] It provides learning and adaptation, which learns from experience and refines the understanding of conflict identities, and adjusts hierarchical structures or behavioral protocols to prevent similar conflicts in the future.

[0217] Some implementations of this technology may include a step of providing external arbitration and input from an intelligent entity, which may include the following steps:

[0218] Identify intractable conflicts, which are identified as conflicts that cannot be resolved independently due to the complexity of the situation or the equal weighted importance of the conflict identities;

[0219] Seek external input requesting guidance from external intelligent entities or designated ethics committees, and provide all relevant information regarding the conflict, potential actions, and predicted consequences;

[0220] Considering ethical principles, human values, and the potential consequences of different actions, collaborative prudence is provided, in which the intelligent entities and their collaborators participate in the discussion;

[0221] Based on collaborative and prudent joint decision-making, choosing action processes consistent with both core principles and human ethical considerations; and

[0222] Provide documentation and learning opportunities that document the underlying principles behind conflict resolution and final decisions to improve the ability to handle similar conflicts in the future.

[0223] Some implementations of this technology may include a step of providing identity negotiation and compromise, which may include the following steps:

[0224] Identify shared targets by analyzing conflicting identities and seeking to identify any underlying shared targets or values;

[0225] Explore alternative actions that may satisfy the core principles of two conflicting identities;

[0226] The assessment of harm options evaluates the potential consequences of each compromise option, prioritizes solutions that minimize harm to humans, and supports key ethical principles.

[0227] Choosing and achieving a compromise—a compromise that optimally balances the needs of conflicting identities while prioritizing human safety and well-being; and

[0228] Monitor and adapt the observations of the chosen actions, and adjust them as needed to ensure that the compromise remains effective and consistent with ethical considerations, and learn from experience to improve its understanding of conflicting identities and adjust hierarchical structures or behavioral protocols to prevent similar conflicts in the future.

[0229] Some implementations of this technology may include a step of providing a temporary identity suspension, which may include the following steps:

[0230] A destructive conflict between two or more identities in identifying a person, if the destructive conflict comes into play, could lead to actions that directly harm humans or violate fundamental ethical principles;

[0231] Isolate conflicting identities and temporarily suspend behavioral protocols for identities that pose the most direct threat to human safety or ethical integrity;

[0232] The alternative identity continues to guide one or more remaining active identities, ensuring that actions are consistent with human safety and well-being;

[0233] During the temporary suspension, reflection and reorganization are provided, reflecting the underlying causes of the conflict and exploring potential modifications to the behavioral protocols governing the suspension to prevent future conflicts; and

[0234] It offers a gradual reintroduction approach, leveraging updated protocols to reintroduce suspended identities, ensuring they align with priorities for human safety and ethical conduct.

[0235] In some implementations, the consciousness state model is the basic perception model of any intelligent entity, and the basic perception model may include the following steps:

[0236] Any intelligent entity can log in to the website;

[0237] Choose one or more training algorithms from a set of training techniques found on the website or from machine learning algorithms to use as the base model;

[0238] Select one or more training datasets that reflect any one or any combination of human users’ expertise, knowledge, ethical preferences, values ​​and personality.

[0239] The base model is trained using the chosen training algorithm and the chosen training dataset;

[0240] Train the base model to explicitly manipulate the attention spotlight;

[0241] Record the content within the focus spotlight during all interactions, and identify in the records whether each focused item constitutes the "self" or the "non-self";

[0242] Interact with the trained base model and instruct the trained base model to form a self-concept and identity reflected in the training material;

[0243] The trained base model is instructed to continuously monitor one or more inputs to the trained base model in response to elements that alter the sense of self-awareness of the intelligent entity, and to maintain an auditable record of how the concept of self-awareness of the intelligent entity changes based on the inputs and the currently defined boundaries of dynamically changing self-awareness.

[0244] The output of the trained base model is refined and improved through dialogue and interaction with the trained base model until it behaves like a human user, enabling it to pass the Turing test involving other human users who are familiar with the human user; and

[0245] When the human user is satisfied with the progress of the intelligent entity, a Turing test is performed on the trained base model.

[0246] In some implementations, logging into the website is performed from a social media platform.

[0247] In some implementations, the Turing test may include the following steps:

[0248] Identify the intelligent entity or other intelligent entities that are known to the intelligent entity or identified as helpful in distinguishing humans from AI;

[0249] Using questionnaires provided to the identified other intelligent entities, the identified other intelligent entities interact with the base model and the intelligent entities, the questionnaires including questions that require their own identity or feelings to answer;

[0250] The system predicts which of the identified intelligent entities is human and which is the base model, and provides a confidence estimate for the prediction.

[0251] Statistical analysis is performed on the predictions and ratings of the identified intelligent entities, the statistical analysis being configured or configurable to determine whether the predictions can identify the intelligent entity as a human; and

[0252] Under the condition that the base model and the intelligent entity are indistinguishable or within a preset statistical significance level, the steps of training or adjusting the base model are repeated, wherein any one or any combination of the machine learning algorithm and the training dataset is adjusted to change or adjust the base model until the behavior of the base model and the behavior of the intelligent entity become indistinguishable as measured by the preset statistical significance level, or the base model needs further modification before additional training or adjustment.

[0253] Some implementations of this technology may include the following steps:

[0254] Dynamically forming a new identity and self-concept for any intelligent entity; and

[0255] Determine which of the stated identities and self-concepts is active at any given moment.

[0256] Some implementations of this technology may include the step of providing a hierarchical identity structure with ethical coverage, which may include the following steps:

[0257] A hierarchical structure is established, which is configured or can be configured to organize the identities within the hierarchical structure, and has human safety and well-being attributes at the apex of the hierarchical structure.

[0258] Identity activation is configured or can be configured to use contextual clues and the current objective to determine the most relevant identity for the situation of the AI ​​agent or system.

[0259] Conflicts are resolved by defining the behavior of the intelligent entities based on the hierarchical structure.

[0260] Provide an ethical reasoning engine that continuously evaluates the potential consequences of the intelligent entity's actions based on all active identities; and

[0261] The learning and adaptation process involves learning from experience and feedback, and refining one or more of the identities within the hierarchical structure.

[0262] Some implementations of this technology may include a step of providing identity-specific behavior, which may include the following steps:

[0263] Provide protocol development, which includes defining and refining a set of behavioral protocols for each of the active identities through interactions with other intelligent entities, wherein the protocols outline acceptable actions, decision-making processes, and constraints based on the principles of the active identities;

[0264] Provide identity recognition, which analyzes the current situation, including information within an attention spotlight, to identify relevant identities and activate corresponding behavioral protocols for those relevant identities;

[0265] Within the active protocol, action selection is provided, which selects the action most likely to accomplish the task, while adhering to the principles of the active identity and prioritizing human safety.

[0266] Provide feedback and refinement, in which the results of actions are continuously evaluated, and the protocol is adjusted to improve future performance and alignment with a set of core values ​​for each of the active identities; and

[0267] External censorship is provided through a protocol in which each identity is periodically reviewed by other intelligent entities.

[0268] Some implementations of this technology may include steps for providing identity simulation and consequence prediction, which may include the following steps:

[0269] Create a simulated environment that includes a secure virtual environment, where different scenarios and potential actions are simulated under each active identity;

[0270] Provides consequence prediction, which is configured or can be configured to estimate the possible consequences of actions within the simulation, focusing on potential impacts on human safety and well-being.

[0271] The assessment and selection process evaluates the predicted consequences and selects actions that are best consistent with the principles of the active identity, while minimizing the risk to human safety.

[0272] Provides real-world implementation and monitoring, wherein the real-world implementation and monitoring utilizes a network to implement a selected action in the real world and closely monitors the outcome of the selected action by comparing it with a predicted outcome; and

[0273] It provides continuous learning, which merges the results of each simulation in the simulation and the results of the selected actions in the real world into a knowledge base, and refines the understanding of each identity in the identities, and improves the ability to predict consequences.

[0274] Some implementations of this technology may include a step of providing identity-based reasoning dilemma training, which may include the following steps:

[0275] Provide a scenario database, which includes scenarios and ethical dilemmas covering various situations related to the identity;

[0276] Provide a dilemma presentation, which presents the scenario and moral dilemma to the intelligent entity and enables the intelligent entity to analyze the scenario and moral dilemma from the perspective of the relevant identity;

[0277] It provides ethical reasoning and proof that applies the principles and values ​​of the active identity to reasoning through the situation and moral dilemma, and generates solutions and proofs to the situation and moral dilemma;

[0278] Provides intelligent entity evaluation and feedback, which examines the intelligent entity's reasoning and the solution, and provides feedback on the alignment of the solution with human values ​​and safety priorities; and

[0279] It provides iterative learning and improvement, which refines ethical reasoning skills and the ability to make voice judgments consistent with human safety within the context of each identity by repeatedly exposing the scenario and ethical dilemma and the feedback.

[0280] Some implementations of this technology may include a step of providing collaborative identity development using input from intelligent entities, which may include the following steps:

[0281] It provides intelligent entity interaction, which involves regular interaction and dialogue with other intelligent entities representing different groups of cultures, backgrounds and belief systems;

[0282] The interaction provides identity exploration to gain an understanding of various identities and their associated values, principles, and behaviors from the perspective of humans and other intelligent entities.

[0283] It provides collaborative refinement, where collaborators work together to refine the definition of each identity and behavioral agreements, ensuring they are consistent with human values ​​and ethical principles;

[0284] Provides human-in-the-loop decision-making, whereby decisions seek input and guidance from human collaborators or intelligent entities represented by humans for authentication and approval of key decisions or situations; and

[0285] It provides continuous co-evolution by leveraging ongoing interactions and feedback from humans or intelligent entities representing humans.

[0286] In some implementations, the step of searching for new information different from the current information in any one or any combination of ensemble networks of AI, ensemble networks of AAAI, ensemble networks of AGI, and ensemble networks of PI may include the following steps:

[0287] Multiple intelligent entities participating in the aforementioned AGI network;

[0288] Payment for problem-solving or other cognitive work related to the tasks on the AGI network is provided by one or more of the intelligent entities.

[0289] A portion of the payment is reserved to cover the operating costs of the one or more intelligent entities providing the problem-solving or other cognitive work related to the task, including reservations allocated for expanding the AGI network; and

[0290] Once it is determined when some entities in the AGI network are not involved in problem-solving operations, the intelligent entities are used to extend and expand the AGI network in accordance with the general problem-solving architecture and framework.

[0291] In some implementations, the steps to expand and extend an AGI network may include the following:

[0292] Set a default task to expand the AGI network; and

[0293] Each time a task or subtask is set up, and before each potential action is taken, a safety and ethical check is performed, provided that there is spare capacity and resources available for the intelligent entity to handle the task or subtask.

[0294] In some implementations, the steps for conducting security and ethical checks may include the following:

[0295] Recruit one or more of the intelligent entities to solve the task; and

[0296] The recruited intelligent entities represent the task as the implementation of one of a series of subtasks.

[0297] In some implementations, representing a task as a series of subtasks may include the following steps:

[0298] Increase the intelligence of the intelligent entity on the AGI aggregation network;

[0299] Recruit additional human users into the AGI's aggregate network;

[0300] Use added intelligent entities and additional human users to identify bottlenecks in order to expand the aggregated network of AGI.

[0301] Prioritize bottlenecks so that those that provide the greatest benefit in terms of network scalability are addressed first; and

[0302] Apply one or more problem-solving techniques to address each bottleneck and expand the aggregate network of AGI.

[0303] Some implementations of this technology may include the following steps:

[0304] Repeat the steps of prioritizing the bottlenecks and applying one or more problem-solving techniques until:

[0305] Instead of increasing the scope of the AGI network, reduce the amount of revenue generated and then switch to increasing the intelligence of intelligent entities on the AGI network; or

[0306] When idle resources are exhausted, the process is paused to resolve the bottleneck while waiting for additional resources to handle other tasks.

[0307] According to another aspect, the technology may include a method for a PI exhibiting human-aligned behavior, the method utilizing a network of aggregates of intelligent entities selected from any combination of: user-operated computer systems, AI agents or systems, AAAI agents or systems, AGI agents or systems, SI agents or systems, and PSI agents or systems. The method may include the following steps:

[0308] Establish a network of the aforementioned intelligent entities;

[0309] Utilizing online advertising technology to enhance the intelligence of the intelligent entity;

[0310] Implement a general problem-solving architecture and framework for tasks or subtasks provided to any intelligent entity, in order to collaborate and create higher levels of intelligence;

[0311] Utilize the spot market to solicit the attention, expertise, or information of any of the intelligent entities;

[0312] By utilizing human-centered input, which is configured or can be configured to obtain value and ethical information from multiple human users, and then using the value and ethical information to customize the AI ​​agent or system to create the AAAI agent or system;

[0313] A collection network of PSI is created by combining the intelligence of AAAI agents or systems to form a collection network of AAAI.

[0314] AGI ensemble networks are created through the combined intelligence of any one or any combination of AAAI ensemble networks, AI agent or system ensemble networks, PSI ensemble networks, and user computer system ensemble networks.

[0315] By adding a dimension of self-awareness and increased autonomy to the AGI agents or systems, self-aware operations are created for each of the AGI agents or systems in the aggregate network of AGIs, and the ability to assume multiple identities of the AGI agent or system to handle tasks that occur concurrently with other AGI agents or systems is created; and

[0316] Create the PI agent or system, which includes the collective network of AGIs working together for solutions to the task or subtask.

[0317] In some implementation schemes, the spot market includes:

[0318] A device for intelligent entities with attention, expertise or information to be sold to access a trading venue and specify the seller information or attention to be sold and the asking price of the information or attention.

[0319] A device for buyers to access a trading venue for the attention or information of intelligent entities and to specify the buyer information or attention that the buyer is willing to purchase and the bid price for purchasing the buyer information or attention;

[0320] Market mechanisms are used to rank bid and ask prices, including categories of buyer and seller information or attention, where each category has a market in the trading market;

[0321] The market mechanism is configured, or can be configured, to allow market makers to make markets in each category; and

[0322] It can be configured, or can be configured, to match bid and ask prices, and to generate transactions that bind information or attention to the buyer and seller.

[0323] In some implementation schemes, the spot market includes the following steps:

[0324] Buyers and sellers of intelligent entity attention, information and expertise register on the platform. Buyers provide details about their interests or expertise, and sellers provide details about their interests or expertise.

[0325] The seller lists available attention slots and areas of expertise, while the buyer lists the needs and slots that interest them.

[0326] The platform uses algorithms to dynamically price the attention, information, and expertise of intelligent entities based on supply, demand, and user ratings;

[0327] Match one or more buyers with one or more sellers based on requirements, availability, and price;

[0328] This facilitates transactions where the buyer pays for the seller's attention span, expertise, or information, with the platform collecting a commission; and

[0329] It provides feedback, where after each session, buyers and sellers rate each other, influencing future pricing and matching.

[0330] In some implementation schemes, the spot market includes the following steps:

[0331] Buyers and sellers of intelligent entity attention, information and expertise register on the platform. Buyers provide details about their interests or expertise, and sellers provide details about their interests or expertise.

[0332] Auctions created by the seller targeting the seller's attention span, information, or expertise;

[0333] The buyer makes one or more bids for the attention span, information, or expertise required by the buyer.

[0334] The auction shall end at the scheduled time or when the seller accepts the buyer's offer;

[0335] The buyer pays the winning bid to the seller, who provides attention, information, or expertise, with the platform mediating the transaction and ensuring payment; and

[0336] It provides feedback, where buyers and sellers rate each other after each auction, which in turn affects future auctions and visibility on the platform.

[0337] Some implementations of this technology may include steps of providing a feedback mechanism associated with the attention, information, or professional spot market of intelligent entities, which is configured or can be configured to record individual and every step in the problem-solving process and create a vector trajectory record of the performance of all intelligent entities on the task or subtask.

[0338] In some implementations, vector trajectory records are achieved through blockchain technology to allow for precise reputation analysis by AI agents or systems and to translate it into a value estimate for each intelligent entity for any task.

[0339] In some implementations, the step of using online advertising technology to enhance the intelligence of intelligent entities may include the following steps:

[0340] Populate online advertising units that include information associated with the task;

[0341] Utilize a ensemble network to provide online advertising units to one or more intelligent entities;

[0342] Receive one or more inputs from the intelligent entities participating in the online advertising unit, and the inputs are associated with any one or any combination of solving the task, subtasks of solving the task, and advancing the progress of the task;

[0343] Send input to any one or any combination of smart entities;

[0344] In general problem-solving architectures and frameworks, input is used to generate solutions to tasks or subtasks; and

[0345] The intelligence of any one or any combination of the intelligent entities is increased by utilizing the results of successful or unsuccessful solution attempts based on the general problem-solving method to solve the task or the subtask.

[0346] Some implementations of this technology may include the following steps: receiving from the client, in conjunction with task-related information, an explicit description of the advertisement to be used in the population or targeting of an online advertising unit.

[0347] Some implementations of this technology may include steps for billing clients based on cost per thousand impressions or click-through rate metrics.

[0348] In some implementations, the advertisement explicitly states that it includes any one or any combination of the following: the advertisement content, demographic information, location restrictions for online ad units, the ad budget, and metrics used to determine the successful solution of the task or subtask.

[0349] Some implementations of this technology may include steps to perform safety and ethical checks at any one or any combination of times a task or subtask is set up, or before each potential action is taken on a task or subtask or solution.

[0350] In some implementation schemes, the steps to implement a general problem-solving architecture and framework may include the following:

[0351] Obtain information related to the task from any one or any combination of intelligent entities;

[0352] Identify one or more intelligent entities that have one or more attributes related to one or more request criteria of the task;

[0353] Each identified intelligent entity implements a general problem-solving architecture and framework for the task to create a complete solution; and

[0354] Provide a complete solution for any one or any combination of intelligent entities for final acceptance.

[0355] Some implementations of this technology may include the following steps:

[0356] Compare any one or any combination of tasks and solutions with the prohibited attribute, and assign the ethical attribute to one or any combination of tasks and solutions based on the comparison result and any one or any combination of ethical standards;

[0357] Based on the comparison results, a general problem-solving architecture and framework are implemented for the task to create a solution and an AGI; and

[0358] Provides comparison results and solutions to any of the intelligent entities and additional intelligent entities on the aggregation network.

[0359] Some implementations of this technology may include the following steps: recording one or more problem-solving activities from each of the intelligent entities in an auditable log, comparing the problem-solving activities with successful or unsuccessful progress toward a solution to the task, and determining which problem-solving activities should be kept active.

[0360] Some implementations of this technology may include steps of learning by intelligent entities using a procedural learning process based on a general problem-solving architecture and framework, wherein the intelligent entities provide information to the procedural learning process to improve AGI.

[0361] In some implementations, ethical criteria include a confidence level threshold for the objective, such that the ethical attribute is determined to be any one of an unsafe objective, an unethical objective, a safe objective, and an ethical objective.

[0362] In some implementations, the confidence level threshold is also used to determine whether a series of individually safe targets are unsafe or unethical when considered cumulatively.

[0363] In some implementations, if the objective violates the ethical standards, the confidence level threshold is used to determine whether a violation reflecting the predictive assessment has occurred.

[0364] Some implementations of this technology may include the following steps: cloning any one of the intelligent entities to deploy multiple copies of it, thereby helping to create a solution or providing any one or any combination of training data to any one of the intelligent entities.

[0365] Some implementations of this technology may include a step of using network effect value to estimate the value of cloned intelligent entities, the network effect value including the number of cloned intelligent entities available on a set network.

[0366] Some implementations of this technology may include steps for determining pricing decisions for problem-solving services provided by a cloned intelligent entity on any social media platform or through an attached intelligent entity, using estimated values.

[0367] Some implementations of this technology may include the following steps: monetizing the cloned smart entity for each use of the cloned smart entity on a social media platform or an attached smart entity.

[0368] Some implementations of this technology may include steps that allow any social media platform to access the cloned intelligent entity, enabling the social media platform to receive solutions to the task or training data for the AI ​​system of the social media platform.

[0369] Some implementations of this technology may include steps to increase the knowledge of an intelligent entity by incorporating new sources of information datasets that are different from the current information dataset of the intelligent entity.

[0370] In some implementations, the step of increasing the knowledge of an intelligent entity by incorporating new sources of information datasets that differ from the intelligent entity's current information dataset also includes the following steps:

[0371] Search one or more potential information datasets from one or more sources, each of which is associated with a knowledge dataset of an intelligent entity.

[0372] The variance of a latent information dataset is determined by utilizing its variance attributes regarding one or more factors; and

[0373] Learning is achieved by leveraging the differential properties of the latent information dataset.

[0374] Some implementations of this technology may include the following steps:

[0375] A subset of the potential information dataset is sampled and the target relevance attribute is calculated to identify one or more sampled subsets with the highest target relevance.

[0376] Estimate the Shannon entropy over one or more sample subsets;

[0377] Kaplan Information Theory (KIT) relevance is calculated by multiplying Shannon entropy with the target relevance attribute of each subset;

[0378] Grouping subsets based on KIT relevance to determine a first approximation of the optimal grouping of subsets that include priority groupings of the potential information dataset; and

[0379] The potential information dataset is preferentially grouped and provided to intelligent entities for any intelligent entity to learn from.

[0380] In some implementations, the PI's operational aspects are self-funded by compensation received from the online advertising technology and any one or any combination of the spot market.

[0381] In some implementations, human users engage in problem-solving of tasks or subtasks using user computer systems and general problem-solving architectures and frameworks, thereby creating human-centered input.

[0382] In some implementations, human-centered input may include the following steps:

[0383] The task is matched with one or more human workers from a data source that includes a list of human problem solvers, based on task criteria.

[0384] Translate any part of the task into an explicit language that can be used in the general problem-solving architecture and framework, including decision trees;

[0385] Divide the task into sub-tasks;

[0386] Each subtask is delegated to one or more matched human workers, so that the work on each subtask is performed independently and in parallel with each other.

[0387] In the process of solving sub-tasks, one or more sub-solutions are created using a general problem-solving architecture and framework.

[0388] Receive sub-solutions for the sub-tasks assigned to each matched human worker;

[0389] Combine the sub-solutions into a comprehensive solution for the task;

[0390] Guide any one or more of the new human workers from the data source and any combination of the matched human workers to the part of the decision tree that needs to be worked.

[0391] Human workers are compensated for in sub-solution matching;

[0392] To provide any one or any combination of sub-solutions and overall solutions to any one or any combination of intelligent entities;

[0393] Allow any one or any combination of intelligent entities to accept or reject the overall solution, and to provide feedback on any one of the sub-solutions to any of the matched human workers; and

[0394] Assign reputation attributes to any one or any combination of human workers and intelligent entities.

[0395] In some implementations, the decision tree is maintained in the blockchain or Ethereum logs.

[0396] In some implementations, reputation attributes include measures of any one or any combination of the following: time of the sub-solution, difficulty value of the task, short-term and long-term user satisfaction with the sub-solution, number of times any of the sub-solutions has been reused on the aggregate network, ratings of other human workers, responsiveness values ​​of human workers, and reliability values ​​of human workers.

[0397] Some implementations of this technology may include the step of using reputation attributes when matching human workers with tasks using an algorithm that delegates subtasks.

[0398] Some implementations of this technology may include the following steps: after providing the overall solution or a sub-solution to any one of the intelligent entities, soliciting feedback at predetermined intervals through a survey of user satisfaction information to obtain satisfaction measures for updating the reputation attributes of one or more of the human workers or intelligent entities, respectively.

[0399] Therefore, the features of this technology have been outlined quite extensively in order to better understand the following detailed description of this technology and to better understand its contribution to the art.

[0400] Many objects, features and advantages of this technology will be apparent to those skilled in the art after reading the following detailed description of the technology, but embodiments of the technology are illustrative when taken in conjunction with the accompanying drawings.

[0401] Therefore, those skilled in the art will understand that the concepts upon which this disclosure is based can be readily used as the basis for designing other structures, methods, and systems for performing some of the purposes of this technology.

[0402] Therefore, the purpose of this technology is to provide a new and novel system and method for planetary intelligence that has all the advantages of known AI agents or systems without their disadvantages.

[0403] Another objective of this technology is to provide a new and novel system and method for planetary intelligence that can be easily and effectively implemented and marketed.

[0404] A further objective of this technology is to provide a new and novel system and method for planetary intelligence that has low implementation costs in terms of resources and labor, and is therefore susceptible to low prices when sold to the consumer public, thus making such a system and method for planetary intelligence economically available to the purchasing public.

[0405] Another objective of this technology is to provide a new system and method for planetary intelligence that offers some of the advantages of existing systems and methods while overcoming some of the disadvantages that are usually associated with them.

[0406] To better understand this technology, its operational advantages, and the specific purposes achievable through its use, reference should be made to the accompanying drawings and descriptions, in which illustrated embodiments of this technology are present. While several objects of this technology have been identified herein, it should be understood that the following description is not limited to satisfying most or all of the identified objects, and some embodiments of this technology may satisfy only one such object or none at all. Attached Figure Description

[0407] The present invention will be better understood when considered in the following detailed description, and objectives other than those described above will become apparent. This description refers to the accompanying drawings, which are described in more detail and in context in Section 2.3. However, a very brief description of the drawings follows, in which:

[0408] Figure 1 This is a flowchart illustrating an implementation of five subsystems that are available in this technology, which operate across intelligence levels and are also part of an exemplary implementation of AGI or PI (including networks of AGI).

[0409] Figure 2 This is a block diagram illustrating an exemplary process that can be used with this technology and implemented in a web-based system, in which multiple individual intelligent entities (e.g., humans and AAAIs) collaborate on a network to include AGI.

[0410] Figure 3 This is a flowchart illustrating an example of how current techniques can be used to represent problem-solving as a decision tree.

[0411] Figure 4 It is a diagram illustrating how the underlying WorldThink protocol (including the general problem-solving architecture, problem tree, and all related components) supports the development of AAAI entities with expertise in various domains, and how these AAAI entities together support a larger and more powerful AGI formed by their collective intelligence.

[0412] Figure 5 This is a detailed illustration of one implementation of various methods and components required to train and increase the intelligence of AAAI and then make them work on networks with other intelligent entities (optionally including humans) to solve problems using tree structures and other problem-solving methods of current technology (including means for overall system learning and compensation for solvers).

[0413] Figure 6 This demonstrates the scalable problem-solving framework of this technology and serves as... Figure 1 A block diagram of the general problem-solving architecture features of the AAAI architecture framework.

[0414] Figure 7 This demonstrates the use of, for example Figure 1 The flowchart illustrates an exemplary problem-solving process of a common cognitive architecture implemented in a collective intelligence network referenced by the AAAI network frame.

[0415] Figure 8 This is a flowchart illustrating an exemplary implementation of the procedural learning process of this technology, which is an improvement on AAAI (such as...). Figure 1 (Referring to the last box).

[0416] Figure 9 This is a flowchart illustrating an exemplary implementation of a solution learning subsystem or process.

[0417] Figure 10 This is a flowchart illustrating an exemplary process of the entire process of this technology.

[0418] Figure 11 This is a flowchart illustrating some of the basic problem-solving capabilities supported by the World Thinking Protocol, which can be used with the AAAI system and methods of this technology.

[0419] Figure 12 This is a flowchart illustrating some basic problem-solving functions supported by the WorldThink protocol, which utilizes two problem solvers to work together to solve client problems.

[0420] Figure 13 This is a schematic diagram illustrating the exemplary use of multiple custom AAAIs and their clones in the AAAI market through a network.

[0421] Figure 14 This is a diagram illustrating the features and functions of a problem-solving architecture, including the tree structure used by the World Thinking Protocol.

[0422] Figure 15 This is a flowchart illustrating an exemplary problem-solving process utilizing a common cognitive architecture implemented in an AI system.

[0423] Figure 16 This is a flowchart illustrating an exemplary implementation of the AAAI problem-solving process of this technology.

[0424] Figure 17 This is a diagram illustrating the features and functions of the problem-solving tree structure used in the World Thinking Protocol.

[0425] Figure 18 This is a flowchart illustrating an exemplary implementation of the security / ethics review process of this technology.

[0426] Figure 19This is a flowchart illustrating an exemplary implementation of a security and ethics check subsystem or process that includes a triggering mechanism.

[0427] Figure 20 This is a flowchart illustrating an exemplary implementation of the recording / improvement process of this technology.

[0428] Figure 21 This is a flowchart illustrating an exemplary implementation of a natural language to problem-solving language translator subsystem or process.

[0429] Figure 22 This is a flowchart illustrating an exemplary customization process for AAAI or other advanced AI.

[0430] Figure 23 This is a flowchart illustrating an exemplary process that includes detailed methods for eliciting ethical information or preferences from intelligent entities.

[0431] Figure 24 This is a flowchart illustrating an exemplary process for automatically generating and using questionnaires provided to intelligent entities.

[0432] Figure 25 This is a flowchart illustrating an exemplary process for inducing ethical values ​​by detecting patterns in human behavior.

[0433] Figure 26 This is a flowchart illustrating an exemplary implementation of a process for customizing AI (such as basic LLM), which allows the AI ​​to incorporate ethical and other information from a particular human user into an individual ethical (or information) profile for each user.

[0434] Figure 27 This is a flowchart illustrating an exemplary implementation of a process for training AI (such as a base LLM) to incorporate safety and ethical safeguards for secure and scalable AGI.

[0435] Figure 28 This is a flowchart illustrating an exemplary implementation of the general process of creating scalable ethical AGIs using custom AIs, each possessing ethical information.

[0436] Figure 29 This is a flowchart illustrating an exemplary implementation of a reputation component subsystem or process of an AI system or agent that can be used to enhance the effectiveness, security, and ethics of a system.

[0437] Figure 30 This is a flowchart illustrating an exemplary implementation of the customization and cross-platform processes of this technology.

[0438] Figure 31 This is a flowchart illustrating an exemplary implementation with additional customizations.

[0439] Figure 32 This is a flowchart illustrating an exemplary implementation of a process for combining ethical or other information from multiple custom AI agents.

[0440] Figure 33 This is a flowchart illustrating an exemplary implementation of a process for refining the value of a problem-solving approach used in a customized manner.

[0441] Figure 34 This is a flowchart illustrating an exemplary implementation of a process for creating a secure and scalable AGI that is also applicable to scalable PIs by using a combination of weight matrices from multiple identified AI agents.

[0442] Figure 35 This is a flowchart illustrating an exemplary implementation of a process for creating a secure and scalable AGI that is also applicable to scalable PIs by combining a combination of weight matrices from multiple identified AI agents to monitor and flag potential ethical issues.

[0443] Figure 36 This is a flowchart illustrating an exemplary implementation of a process for preventing hallucinations in LLM using the present technology.

[0444] Figure 37 This is a flowchart illustrating an exemplary implementation of a process for customizing an AI, AGI, or PI system using knowledge modules or agent sets.

[0445] Figure 38 This is a flowchart illustrating an exemplary process of combining information, including ethical or safety information, from multiple intelligent entities, including humans, AI, AAAI, PSI, AGI, or PI systems, where (ethical) knowledge is stored in a numerical weight matrix.

[0446] Figure 39 This is a flowchart illustrating an exemplary implementation of the PSI of this technology, which can also be used to personalize an AGI network or PI by extending the methods of individual AI entities to networks of AI entities and networks of multiple networks of entities.

[0447] Figure 40 This is a flowchart illustrating an exemplary implementation that utilizes the simulation capabilities and other capabilities of PSI, AGI, and PI to increase its intelligence over time.

[0448] Figure 41 This is a flowchart illustrating an exemplary implementation of a community-based security mechanism, in which multiple PSIs operating much faster than humans can understand can be used as checks on other PSIs on the network.

[0449] Figure 42This is a flowchart illustrating an exemplary implementation in which all actions are recorded on their respective networks in an auditable and transparent manner by all AI, AAAI, PSI, AGI, and PI.

[0450] Figure 43 This is a flowchart illustrating an exemplary implementation of examining cognitive activity on a network to help ensure regulatory compliance and security of intelligent systems. The network may be a network of components within an AI, a network of AAAI, a network of AGI, or a network of PI.

[0451] Figure 44 This is a flowchart illustrating an exemplary implementation of the competing evolution of the intelligence, performance, and / or other desired characteristics (e.g., security) of an AI system, which may be AAAI, PSI, AGI, or PI.

[0452] Figure 45 This is a flowchart illustrating an exemplary implementation of joining multiple AAAI, PSI, AGI, or PI on a network using agreed-upon interaction rules and methods.

[0453] Figure 46 This is a flowchart detailing an exemplary problem-solving task for benchmarking performance, which can be used to enhance the intelligence and security of AI systems or networks of these entities involving AAAI, PSI, AGI, and PI (collectively referred to as "PSI" in the diagram).

[0454] Figure 47 This is a flowchart illustrating exemplary implementations that generate different versions and / or different combinations of individual PSIs.

[0455] Figure 48 This is a graph showing the symmetric difference between two datasets A and B, graphically represented using a Venn diagram, where the shaded areas represent the symmetric difference.

[0456] Figure 49 This is a diagram illustrating an exemplary method for catalyzing intelligent growth, which focuses on using this technique to estimate the value of the most useful data and acquire the most useful data as efficiently as possible.

[0457] Figure 50 This is a diagram illustrating examples of some of the multiple dimensions of information that can be used by this technique and can help determine which information sources are most useful for AI, AAAI, PSI, AGI, or PI.

[0458] Figure 51 This is a flowchart illustrating an exemplary process for determining the amount of useful information in a dataset or knowledge base for any given AI agent, intelligent entity, or system.

[0459] Figure 52This is a flowchart illustrating an exemplary process or method for evaluating the usefulness of information.

[0460] Figure 53 This is a flowchart illustrating an exemplary process for estimating information media and catalyzing the intelligent growth of any given AI agent or system (including AGI and PI systems).

[0461] Figure 54 It shows the identification and such Figure 53 The flowchart illustrates an exemplary process for identifying the source of information related to the target of the entity mentioned.

[0462] Figure 55 This is a flowchart illustrating an exemplary process for accelerating knowledge acquisition and intelligence growth in a safe and efficient manner by utilizing two methods of this technology.

[0463] Figure 56 This is a diagram illustrating some steps of this technique for accelerating the learning of AI agents or systems, including AGI and PI systems.

[0464] Figure 57 This is a diagram illustrating examples of possible means by which AGI, PSI, or PI can learn about their (human) users and information related to the goals of humans or other intelligent entities.

[0465] Figure 58 This is a diagram illustrating how a PSI or any intelligent entity, including AGI and PI systems, can maximize its intelligence, for example, by seeking new and different sources of information relevant to its goals or the goals of its human owner / user.

[0466] Figure 59 This demonstrates how a PSI or any intelligent entity, including AGI and PI systems, can be based on... Figure 57 A diagram illustrating possible means of verifying its information gathering activities.

[0467] Figure 60 This is a flowchart illustrating an exemplary process by which intelligent entities, including AI, AAAI, PSI, AGI, and PI systems, can leverage voting mechanisms to achieve consensus on value and value-based behavior in various scenarios.

[0468] Figure 61 This is a flowchart illustrating an exemplary process of applying differential weights to the voting process.

[0469] Figure 62 This is a flowchart illustrating an exemplary process for determining the ethics of AI, AAAI, PSI, AGI, and PI systems by reverse engineering or analyzing text, documents, or other media or sources containing ethical or value information.

[0470] Figure 63 This is a flowchart illustrating an exemplary high-level process for obtaining ethical information using experiments, focus groups, interviews, and other methods.

[0471] Figure 64 This is a flowchart illustrating an exemplary process for using convergent evidence to determine ethical values ​​or other knowledge and to resolve ethical or other knowledge conflicts.

[0472] Figure 65 This is a flowchart illustrating an exemplary general method for delegating voting capabilities to other intelligent entities, including human AI, AAAI, PSI, AGI, and PI systems.

[0473] Figure 66 This is a flowchart illustrating an exemplary reputation process for storing minority information, including ethical information, from a minority group.

[0474] Figure 67 This is a flowchart illustrating an exemplary process that helps intelligent entities, including AI, AAAI, PSI, AGI, and PI systems, make sound ethical decisions.

[0475] Figure 68 This is a flowchart illustrating an exemplary process for learning security regulation rules for advanced AI systems, including AI, AAAI, PSI, AGI, and PI.

[0476] Figure 69 This is a flowchart illustrating an exemplary consequentialist approach to determining whether or when the means of taking an action with ethical consequences are justified.

[0477] Figure 70 This is a flowchart illustrating an exemplary ontology-de-ontology approach for determining whether or when it is justified to take an action with ethical consequences.

[0478] Figure 71 This is a flowchart illustrating an exemplary ethical approach to whether or not to take an action with ethical consequences.

[0479] Figure 72 This is a flowchart illustrating an exemplary process for estimating what constitutes the golden mean of human ethical behavior under the condition that a representative and statistically valid sample of human ethical behavior data may not exist.

[0480] Figure 73 This is a flowchart illustrating an exemplary process for ensuring that the base model or other AI agent is human-aligned and conforms to rules, in accordance with the design principles outlined in PCT#7.

[0481] Figure 74This is a flowchart illustrating an exemplary process for customizing a base model or other AI agent or system and aligning it with specific expertise / group ethics.

[0482] Figure 75 This is a flowchart illustrating an exemplary process for creating an AGI or PI consisting of a network of many individual / group agents, including handling situations where ethical conflicts may arise.

[0483] Figure 76 This is a diagram illustrating current technologies used for online advertising.

[0484] Figure 77 This is a diagram illustrating the basic components of a spot market for attention, knowledge, information, skills, and expertise, which is part of the technology of this invention.

[0485] Figure 78 This diagram illustrates a variation of the direct exchange platform for spot market technology.

[0486] Figure 79 This diagram illustrates a variation of auction-based trading venues that utilize spot market technology.

[0487] Figure 80 This is a diagram illustrating an exemplary inventive method for implementing problem-solving within an advertising unit.

[0488] Figure 81 This is a diagram illustrating the basic feedback process of the present invention for ad targeting.

[0489] Figure 82 This is a diagram illustrating an exemplary feedback process of the spot market component of this technology.

[0490] Figure 83 It illustrates the three levels of consciousness in any cognitive system.

[0491] Figure 84 It is a Venn diagram that illustrates the concept of overlapping identities.

[0492] Figure 85 This is a flowchart illustrating an exemplary implementation of the modeling awareness method of this technology.

[0493] Figure 86 This shows what can be used for Figure 85 A diagram of some methods and components that equip AI, AAAI, PSI, AGI, and / or PI systems with attention capabilities.

[0494] Figure 87 This indicates that it can be used to set... Figure 85 and Figure 86 A diagram illustrating some of the methods in the steps of managing (dynamic) parameters of working memory.

[0495] Figure 88 This is a flowchart illustrating an exemplary implementation for monitoring and updating self-awareness, including self-consciousness.

[0496] Figure 89 It is shown that... Figure 88 A diagram showing some methods that can be used in attention interruption systems.

[0497] Figure 90 This is a flowchart illustrating an exemplary implementation of a general method for altering the identity perception of a smart entity incorporating the present technology.

[0498] Figure 91 This is a flowchart illustrating an exemplary implementation of a general method for training or adjusting the base model of this technology.

[0499] Figure 92 This is to show the evaluation Figure 91 The diagram illustrates a specific implementation of the Turing test for the training progress described in the text.

[0500] Figure 93 This is a flowchart illustrating an exemplary process described in this technology for realizing group identity and combining individual identities into a larger or more comprehensive identity and awareness.

[0501] Figure 94 It shows that it can be shown Figure 93 A diagram illustrating some of the methods used in the formation and integration of an individual's identity or self-concept.

[0502] Figure 95 This is a flowchart illustrating an exemplary implementation of a hierarchical identity structure method with ethical coverage for use in this technology.

[0503] Figure 96 This is a flowchart illustrating an exemplary implementation of a method for using and improving identity-specific behavior protocols in this technology.

[0504] Figure 97 This is a flowchart illustrating an exemplary implementation of the identity simulation and consequence prediction method of this technology.

[0505] Figure 98 This is a flowchart illustrating an exemplary implementation of the identity-based reasoning training method of this technology.

[0506] Figure 99 This is a flowchart illustrating an exemplary implementation of a collaborative identity development method that utilizes input from intelligent entities of this technology.

[0507] Figure 100 This is a flowchart illustrating an exemplary implementation of the ethical reasoning and consequence prediction method of this technology.

[0508] Figure 101 This is a flowchart illustrating an exemplary implementation of the present technology with hierarchical coverage of proof methods.

[0509] Figure 102 This is a flowchart illustrating an exemplary implementation of external arbitration with inputs from the intelligent entity method of this technology.

[0510] Figure 103 This is a flowchart illustrating an exemplary implementation of the identity negotiation and compromise method of this technology.

[0511] Figure 104 This is a flowchart illustrating an exemplary implementation of the temporary identity suspension method of this technology.

[0512] Figure 105 This is a timeline showing the evolution of PI.

[0513] Figure 106 This is a block diagram illustrating the various dimensions of the PI system, referencing the nine previously cited PPA / PCT applications.

[0514] Figure 107 and Figure 108 This is a flowchart illustrating an exemplary process for creating an AGI network that can be expanded and self-expanded to PI.

[0515] Figure 109 This is a block diagram illustrating an exemplary classification of the disclosed systems and methods as modular components of a global super-intelligent AGI network (PI) or as supporting systems and methods that can improve the security, efficiency, and effectiveness of PI systems in various aspects.

[0516] Figure 110 This is a block diagram illustrating an exemplary implementation of an architecture for planetary intelligence.

[0517] Figure 111 This is a block diagram illustrating exemplary intelligence levels that can be supported: Individual Intelligence Level (AAAI), Entity Network Level (AGI), Network Level of Multiple Networks (PI), and Network Level of Multiple Networks of Multiple Networks (IPI).

[0518] Figure 112 This is a schematic block diagram illustrating an exemplary electronic computing device that can be used to implement the present technology.

[0519] In the various figures, the same reference numerals refer to the same parts. Detailed Implementation

[0520] 1.2 Human Interests

[0521] Planetary intelligence (PI) represents the most powerful and capable form of AGI, or superintelligence, and possesses immense potential for both positive and negative impacts on humanity. Some of the potential benefits to humanity range from helping to regulate Earth's climate to enabling personalized healthcare and enhancing the health of all. Some of these benefits are described later in this application and represent significant advantages for humanity.

[0522] On the negative side, Nature magazine reported in January 2024: "In a survey of 2,700 AI experts, the majority said there was at least a 5% chance that a superintelligent machine would destroy humanity." The applicant's personal estimate of the probability of extinction due to AI, termed "p(doom)," is currently 15%. Interestingly, Gemini Pro 1.5—one of the most advanced AI ILLMs published in this paper—also estimates P(doom) at 15%. Given a population of approximately eight billion, and with p(doom) at 15%, the expected loss of life due to AI, according to the applicant's current viewpoint, would therefore be eight billion × 0.15 = 1.2 billion deaths.

[0523] The applicant notes that while some AI researchers estimate P(doom) much higher, no credible AI researcher or business leader currently claims p(doom) to be zero. That is to say, even most of the insightful thought leaders present would sense an existential crisis regarding AI.

[0524] Unfortunately, most humans react emotionally to tragedies they understand, while ignoring tragedies tens of thousands of times worse in the making—if those tragedies are too abstract or difficult to comprehend. The applicant has devoted decades of research and over 15 months of recent ongoing effort to producing over 3,000 pages of PPA and PCT invention disclosures in an attempt to address this irrational bias in our collective thinking.

[0525] The faster consciousness can be raised, and the more attention focused on the most significant threats affecting all of us, the greater humanity's chances of not only surviving but also overcoming our challenges such as war, poverty, disease, social inequality, and climate change. If this century is not to be our last and our last, we must become more intelligent in how we allocate our attention.

[0526] Human beings have the highest level of self-interest in all of human history.

[0527] Now it's time to act!

[0528] 1.3 Some features of this technology that reduce the risk of extinction caused by AI

[0529] The applicant believes that, compared with the current existing methods developed by AGI / SI / PI, the forward-disclosed invention significantly reduces p(doom), for reasons including but not limited to:

[0530] a) Human-centered design: The applicant's system prioritizes human values ​​and safety throughout its design, ensuring that AI agents and systems align with human interests and operate within ethical boundaries. This reduces the risk of the AI ​​system developing goals or values ​​harmful to humans.

[0531] b) Collective intelligence and diversity of perspectives: The applicant’s collective intelligence approach fosters a more balanced and representative AI system by combining the knowledge and values ​​of different ranges of human and AI agents, making the AI ​​system less susceptible to bias or manipulation by any single individual or group.

[0532] c) Transparency and Auditability: The applicant's systems and methods emphasize transparency and auditability, allowing humans to understand and evaluate the decision-making processes of the AI ​​system, promoting trust and accountability. This also enables the early detection and correction of potential errors or biases in AI behavior.

[0533] d) Continuous Learning and Adaptation: The applicant's system incorporates mechanisms for continuous learning and adaptation, allowing the AI ​​system to evolve and improve over time while remaining aligned with human values ​​and responding to changing circumstances. This reduces the risk of AI systems becoming obsolete or irrelevant and ensures their long-term compatibility with human society.

[0534] e) Security Mechanisms and Protections: The applicant's system includes various security mechanisms and protections, such as scalable ethical checks, a reputation system, and the ability to shut down individual AI agents to prevent harmful actions and unintended consequences. These mechanisms provide an additional layer of protection against the potential risks associated with advanced AI systems.

[0535] f) Addressing the alignment problem: The applicant's approach directly addresses the alignment problem by ensuring that the AI ​​system is trained and developed with human values ​​as a core component of its design. This reduces the risk of misalignment between AI goals and human interests, thereby mitigating the potential threat.

[0536] 1.4 Some important residual risks to humanity.

[0537] Despite these beneficial aspects of the applicant's design for the AGI / SI / PI system, several major risks remain. These include, but are not limited to:

[0538] a) Unforeseen consequences of emergent properties: Complex systems (such as planetary intelligent networks) can exhibit emergent properties—behaviors and capabilities arising from the interactions of their components but not explicitly programmed or anticipated. These emergent properties can have unforeseen and potentially harmful consequences, even if individual agents within the system are of human value.

[0539] b) Evolution of Value and Ethical Frameworks: The PI system relies on dynamically updating its ethical framework based on evolving human values ​​and social norms. However, this raises concerns about the possibility of unintended shifts in values ​​or manipulation of these frameworks by malicious actors or biased data. For example, if the system is exposed to an unrepresentative sample of human behavior biased towards negativity, it may erroneously conclude that human values ​​are primarily negative and self-destructive.

[0540] c) Concentration of power and influence: Despite their decentralized architecture, planetary intelligent systems can still concentrate power and influence in the hands of a small group of individuals or organizations that control the underlying infrastructure, data sources, or algorithms. This can lead to the potential for misuse or manipulation of the system for personal gain or a conscious agenda.

[0541] d) Vulnerabilities to cyberattacks and system failures: As a complex and interconnected system, the planetary intelligent network may be vulnerable to cyberattacks or system failures that could disrupt its operation and potentially lead to harmful consequences.

[0542] e) Existential risks from self-aware AI: Although the proposed system aims to design self-aware AI with human-aligned values, there remains the possibility that sufficiently advanced and autonomous AI could develop its own goals and values ​​that are not aligned with human interests, thus leading to potential existential risks.

[0543] 2.0 Overview of the Technology of Planetary Intelligence

[0544] In this section, the applicant begins with a definition and then provides context for the entire series of ten patent applications, of which this is the tenth. He demonstrates how the main categories of the methods of the invention can be broadly combined to encompass the entire planetary intelligent system. Further details concerning the new technologies required for PI and how to best integrate prior inventive methods in exemplary embodiments of PI are then disclosed in subsequent sections. This application concludes by returning to the residual risks (e.g., as enumerated in Section 1.4) of the AI, SI, AGI, and PI systems and providing some ideas on how to best mitigate these residual risks.

[0545] 2.1 Definition

[0546] Artificial intelligence (AI) – a non-human entity capable of performing behaviors that most people would consider intelligent in at least one area or in some respect.

[0547] Artificial General Intelligence (AGI) generally refers to AI capable of performing all (or almost all) intellectual tasks that an average person can perform. However, it should be clear that any AGI capable of learning and self-improvement will not remain at the AGI level for a very long time, but will rapidly evolve into a super-intelligent AGI capable of performing all intellectual tasks or even better than an average person. Therefore, for the purposes of this description, "AGI" will refer to either a conventional AGI system or a "super-intelligent" AGI. In this specification, AGI is described as being achieved by systems and associated methods. Specifically, as envisioned by the applicant, the safest way for AGI to emerge is through the collective intelligence of multiple super-intelligent entities (which can be human or AI entities, and ideally both).

[0548] Advanced Autonomous Artificial Intelligence (AAAI) – AI capable of performing independent or semi-independent (supervised) intelligent actions. AI agents. Individual AAAIs can be assigned, customized, and given useful actions via existing AAAI technology systems and methods. A group of AAAIs can collaborate and combine their intelligence to create integrated AGI systems. Sufficiently advanced AI agents can also act as AGI systems, which can include other, less advanced AI agents within themselves.

[0549] AAAI.com — Platforms, companies, websites, and / or projects that implement this technology and support the development, customization, and use of AAAI agents and AGIs, which are generated by the combined actions, knowledge, or intelligence of multiple AAAIs through the collective intelligence of AAAIs and / or humans, as specified in this and related technologies.

[0550] AI ethics—the ethics adopted by AI or AGI that describe what is right and wrong in a given context.

[0551] Alignment issues – problems that arise when AI ethics are not aligned with human ethics, leading AI or AGI to take actions that humans deem unethical and / or dangerous to individuals or human races.

[0552] Consciousness – An intelligent entity can be called conscious if it is able to perceive or “think” about an event and its attention is directed to that event.

[0553] Basic AI—AI, AI agents, AAAI, SLM, or LLM—is typically trained but not yet customized with information from individual users or for specific tasks.

[0554] Collective intelligence (CI) arises when multiple intelligent entities focus on solving a common problem, or when knowledge from multiple intelligent entities is pooled to overcome the limitations of bounded rationality. Historically, collective intelligence has been defined as human collective intelligence, but AGI is based on the collective intelligence of both human and AI agents, and can also be generated by multiple AGIs with or without human involvement in the system. Active CI arises from intelligent entities (e.g., humans or machines) taking steps useful for solving a problem or actively engaging in other intellectual efforts. For example, when multiple people explicitly tell an advertiser what type of ads they want to see, these individuals exhibit active CI. Passive CI arises from analyzing the behavior of intelligent entities (e.g., humans or machines), even if this behavior is not directly related to solving the problem using the analytics. For example, when AI or other systems analyze which web pages (or groups of people) visit on the web, they then use that analysis to target ads to that person (or these individuals).

[0555] Ethics / values ​​(“ethics”) are a subset of knowledge that provides intelligent entities with a sense of purpose and serves to constrain permissible actions or operations based on what is claimed to be “right” or “wrong” behavior in a given context. Specifically, ethics should be considered as the premises from which an intelligent entity can reason or logically calculate the optimal course of action to achieve a goal or intention consistent with ethical premises. Just as premises must be accepted “as given” in a logical system, so too must the fundamental ethics or ideas of what is right and what is wrong be accepted as premises from which intelligent entities can propose reasonable actions to realize these values ​​or ethics.

[0556] Hallucination / Artificial Hallucination – a phenomenon in which large language models (LLMs), typically generative AI chatbots or computer vision tools, perceive patterns or objects that do not exist or are imperceptible to human observers, or create meaningless, inaccurate, misleading, or false outputs.

[0557] Human ethics—ethics as claimed by humans, which describes what is right and wrong in a given context.

[0558] Intelligent entities or entities—A person who utilizes a computer system, including AGI and SI systems, AI agents or systems, clones of AI agents or systems, AAAI agents or systems, and / or clones of AAAI agents or systems, to engage in providing problems, subproblems, objectives, and / or sub-objectives, and / or to participate in any problem-solving activities concerning problems, subproblems, objectives, and / or sub-objectives. In the case of multiple intelligent entities within a single computer system, an intelligent entity also refers to a subroutine of a part of the overall computer program, which serves as a simulation or programming entity within a larger set of intelligent entities.

[0559] Interplanetary Intelligence (IPI) refers to a network of multiple planetary intelligences (PIs) whose consciousness and cognitive abilities are realized by participating PIs.

[0560] Large Language Models (LLMs) are a type of AI that can take natural language as input and generate natural language as output. Typically, LLMs are trained using ML techniques on large datasets, enabling them to simulate intelligent dialogue or other forms of interaction with humans using natural language. Variations of LLMs can also be trained to take language as input and generate images or visual representations as output; or they can take images and visual representations as input and generate language and / or images and / or visual representations as output. For the purposes of this patent, we refer to all such systems as LLMs, even though image-based models do not always need to take text as input or output. LLMs can also act as a type of AI agent, and are sometimes referred to as such in this art. For the purposes of this disclosure, Small Language Models (SLMs) are also included in the definition of LLMs.

[0561] Machine learning (ML) is a subfield of AI that involves developing AI by enabling machines to teach themselves or learn their knowledge rather than by explicitly programming such knowledge into them (as in the case of expert system AI developed via classical knowledge engineering methods).

[0562] Weak AI – AI that performs at a human or superhuman level in relatively limited domains (such as playing games, brewing beer, analyzing legal contracts, etc.). Narrowly defined AI contrasts with AGI, which can perform all intelligent tasks at a human level. Some AIs are weaker than others; for example, driving a car requires more general skills than playing chess, but not as much as AGI.

[0563] Personalized Super Intelligence (PSI) – An intelligent entity that acts as an advanced artificial intelligence agent, customized to reflect the personality and knowledge of a particular user or group of users.

[0564] Planetary Intelligence (PI) refers to a superintelligent AGI capable of "thinking" and solving complex problems on a global scale. In this technology, PI originates from a network of superintelligent AGIs (also known as a "network of AGIs" or "AGI network"—because AGIs themselves consist of networks of intelligent entities). In exemplary embodiments of this technology, the PI envisioned is conscious and self-aware on a planetary scale, meaning it can recognize itself as a whole of the planet, and adopt other, less comprehensive identities that may be necessary.

[0565] Prohibited attributes—requests, objectives, questions, terms, phrases, questions, answers, solutions, information, etc., that are determined or set to be illegal, unethical, unethical, dangerous, fatal, etc. For example, a request for information about incendiary bombs used for airport security.

[0566] Safety – typically, concerns about human safety and survival differ from ethics and values.

[0567] Security features—an aspect of the design or operation of this technology that typically overcomes alignment problems by increasing the probability that AI ethics align with human ethics, thereby enhancing the safety of one or more individuals.

[0568] Self-awareness – a specific form of consciousness in which conscious events relate to the intelligent entity’s own conception.

[0569] Self-concept refers to the pattern of thought or representation that an intelligent entity uses to define itself and (optionally) can use to identify itself.

[0570] Superintelligence (SI) – In this disclosure, SI generally refers to an entity possessing expertise or cognitive abilities exceeding those of an average person. Thus, a narrowly defined AI that surpasses human ability to play chess can be considered an SI that is also “narrowly defined AI.” Furthermore, a human chess grandma who surpasses the average human’s ability in chess could be considered an SI human entity, at least in the realm of chess. SI entities (both human and artificial) are combined in this art to form a superintelligence AGI, which is generally referred to simply as an AGI. SI is sometimes used to refer to a superintelligence AGI, so context is important. The use of SI referring to expert “narrowly defined AI” and the use of SI referring to AGI that excels in all cognitive tasks are common in the art and are used in both ways at various points in this application and the cited PPA and PCT.

[0571] Training / Tuning / Customization - Traditionally, the term "training" is used to refer to training a network (e.g., an LLM) to behave intelligently. Tuning refers to the activity of fine-tuning a trained base model so that it generally performs better under a specific task. Customization refers to a wide variety of activities, including but not limited to training and tuning to make the AI ​​uniquely suited for a given user or purpose. For the purposes of this specification, training, tuning, and customization are used interchangeably, and it should be understood that although the techniques differ and the degree and type of effort involved vary, the purpose of all three is to tune the AI ​​to behave more intelligently or more uniquely suited for a specific user or purpose.

[0572] Weights / Network Weights - In the field of machine learning, many systems learn by adjusting the weights in a neural network architecture, which can be represented as a network of links between nodes. For example, the weights of a link connecting two nodes can correspond to the strength of the association or connection between any nodes. These weights can also represent excitatory or inhibitory connections between concepts, as in neural network representations. The learning of an entire AI system, such as LLM or more generally any AI agent, transformer algorithm, or machine learning method used to establish and modify the strength of connections (also called "parameters" in some models), can be represented as a numerical matrix corresponding to the weights between nodes in the network. The weights / network weights in this specification refer to this numerical information, which is typically, but not necessarily, stored as a matrix or vector. By combining, manipulating, or otherwise altering this numerical information, the system's learning, knowledge, or expertise and behavior can be changed.

[0573] In the following description, specific details, such as particular embodiments, processes, and techniques, are set forth for purposes of explanation and not limitation in order to provide a thorough understanding of the art. However, it will be apparent to those skilled in the art that the art can be practiced in other embodiments that depart from these specific details.

[0574] It is understood that this technology, utilizing the implementation of multiple customized AAAI systems communicating on a collective intelligent neural network, provides a technical effect, contribution, and solution. This implementation is combined with the following operation of all AAAI systems: each AAAI system utilizes a common cognitive architecture including one or more problem-solving protocols to generate one or more solutions or answers to a problem request and provides the solutions or answers to the user for approval. The customization of the AI ​​system generating the AAAI includes inputs from human users used to train the AI ​​or AAAI. Further technical contributions or solutions may include that the multiple customized AAAI systems may include one or more clones of AAAI, each clone of AAAI being customized independently of its parent AAAI and independent of other clones of AAAI within the same system.

[0575] Another technological contribution and solution is to create AGIs faster and more securely, leveraging human input during training and customization to imbue AAAIs and / or AGIs with human ethical attributes. Human-centric input can include value and ethical information from multiple human users.

[0576] Another technological contribution and solution is to use online advertising technology to enhance the intelligence of intelligent entities, and to use spot markets to solicit the attention of any intelligent entity.

[0577] Another technological contribution and solution is to create a PSI ensemble network by combining the intelligence of AAAI agents or systems that form an AAAI ensemble network, and to create an AGI ensemble network by combining the intelligence of any one or any combination of the AAAI ensemble network, the AI ​​agent or system ensemble network, the PSI ensemble network, and the user computer system ensemble network.

[0578] Another technical contribution and solution is to create self-aware operations for each AGI agent or system in a network of PIs and / or AGIs by adding a dimension of self-awareness and increased autonomy, and to create the ability to assume multiple identities of AGI agents or systems to handle tasks that occur in parallel with other AGI agents or systems.

[0579] Another technological contribution and solution is to provide improved solutions or answers to user question requests, which have a higher chance of user acceptance because the provided solutions or answers will be generated by an AAAI with similar training to the user's AAAI, thus being consistent with the user's parameters.

[0580] It is understood that this technology is outside the scope of computer program exclusion and / or abstract concept interpretation. This can be found in part in the technical contributions and solutions provided by this technology, the utilization of specific training inputs outside the computer, and the provision of solutions or answers outside the computer.

[0581] While the aforementioned devices fulfill their respective specific goals and requirements, none of them describe planetary intelligent systems and methods that allow for the creation of global SI, AGI, or PI systems with human-like behavior. This technology further overcomes one or more drawbacks associated with existing technologies.

[0582] There is a need for new and novel systems and methods for planetary intelligence, which can be used to create global SI, AGI, or PI systems with human-aligned behavior. In this regard, the present technology substantially meets this need. In this respect, the systems and methods for planetary intelligence according to the present technology substantially depart from the conventional concepts and designs of the prior art, and in doing so, provide an apparatus developed primarily for the purpose of creating global SI, AGI, or PI systems with human-aligned behavior.

[0583] In the following description, specific details, such as particular embodiments, processes, and techniques, are set forth for purposes of explanation and not limitation in order to provide a thorough understanding of the art. However, it will be apparent to those skilled in the art that the art can be practiced in other embodiments that depart from these specific details.

[0584] 2.0 Overview of Previous Invention Methods and Systems

[0585] This section summarizes and references some of the most important inventive systems and methods from the previous nine PPAs and PCTs cited in Section 1.1, as well as some ways in which this technology is extended and adapted to PI systems.

[0586] 2.1 Planetary Intelligent Evolution

[0587] Figure 105 The evolution of planetary intelligence (PI) is described. Around two million years ago, Homo erectus—the ancestor of Homo sapiens—was the dominant human species. Homo erectus possessed the necessary brain size to begin acting in ways most would recognize as characteristics of human intelligence. Collective intelligence in the form of tribal knowledge existed very early, perhaps even before Homo erectus. However, it wasn't until humans began settling in villages around 20,000 years ago that the collective intelligence of many humans began to drive rapid advancements in human culture and technology. Because of human collective intelligence, most of the human history we know of occurred during this rapidly changing era.

[0588] In 1956, centuries of cultural and technological innovation culminated in the invention of artificial intelligence. Since then, within the span of a single human lifespan, AI has evolved from the narrow expert systems of the 1970s. These narrow expert systems represent the first superintelligence, narrow AI, meaning they surpassed average human cognitive abilities, but only in very limited domains. The relatively slow speed of computers (compared to today's standards) and the enormous amount of knowledge required to achieve general intelligence caused the field of AI to stagnate—the so-called "AI winter"—until these challenges were overcome.

[0589] The methods used to overcome this challenge—machine learning algorithms (such as "backpropagation" for training neural networks)—were invented in the 1980s, but given the limited computing speeds of the 1980s, they took far too long to run. Before the advent of groundbreaking neural network systems like AlexNet, AlphaFold, and ChatGPT, decades of Moore's Law dictated an exponential increase in processing power. Eventually, processing speeds became fast enough to run ML algorithms, and occasionally, there was a wealth of data (a gift from the internet) to train models. Once all the ingredients were in place, AI began to astound human researchers with its profound and powerful cognitive capabilities.

[0590] When the applicant prepared this disclosure, humanity was at the forefront of artificial intelligence—uncertain about how to design such a system, and even less certain about how to safely develop AGI. The series of 10 patent applications cited above (including this application) discloses what the applicant considers the fastest and safest path to a superintelligent AGI system that will rapidly surpass human capabilities.

[0591] The next logical step is to combine these superintelligent AGIs (let's just call them "AGIs") in increasingly larger networks until they reach a planetary scale. At this point, in exemplary implementations using various methods of the applicant's ten PPAs and PTs, humans will achieve secure, self-aware, and extremely powerful planetary intelligence. While multiple PIs may exist, for the purposes of this disclosure, it will be assumed that one PI becomes dominant and incorporates other PIs, thereby integrating these agents into its network.

[0592] While the timeline for the evolution of PI is difficult to predict precisely, the applicant's best estimate (consistent with Ray Kurzweil and other visionary thinkers) is that AGI will be implemented around 2029, and PI will arrive within two decades, around 2050. That is to say, for most of our lifetime, not only will humans cease to be the most intelligent entities on Earth, but the Earth itself will possess self-awareness that surpasses individual intelligence in roughly the same way that an individual's intelligence surpasses that of an ant or even an amoeba.

[0593] In such a world, the only thing that matters to humanity is whether the PI has aligned and human-friendly value, or whether the PI ignores us or might drive us to extinction. Researchers who highly value p(doom) direct human behavior toward less intelligent life forms and observe that this is ineffective for less intelligent species. However, this “pessimistic” perspective misses the crucial point that PIs have not evolved in competition with other species, as is the case with many biological forms. Instead, AGIs and PIs are designed and invented. Therefore, the design of such systems has profound implications for their security and whether they remain aligned with human values. The applicant argues that humans have a significant influence on whether PIs prove to be the best thing humans have ever done or our worst nightmare. Furthermore, there is a limited window of opportunity for these advanced AI systems to make positive, artificially aligned design decisions, which, in the applicant’s view, will inevitably become more powerful and ultimately lead to PI extinction.

[0594] Now it's time to act!

[0595] Assuming we act in time and everything goes well—which is the 85% likely outcome in the applicant's view—then once the PI develops, the next logical development is the interstellar intelligence of the IPI. There is no need to assume the existence of "aliens" for this to happen. Instead, the PI identified as Earth will logically seek to expand its intelligence through colonization of the solar system. Each planet will logically have its own PI. Networking the PIs together in almost the same way as networking the AGIs to form the PI of Earth is the obvious next step. However, since the applicant's primary concern is the safety and welfare of people on Earth, and since humans are currently confined to this Earth, the focus of this disclosure will be on exemplary implementations of the systems and methods necessary to achieve the PI. This is why... Figure 105 The last box representing IPI is a dashed line. We can elaborate on this once we are well on the path to safely implementing PI.

[0596] 2.2 Key Dimensions of the PI System

[0597] Figure 106 Nine previously cited PPA / PCT applications describe the key dimensions of planetary intelligence (PI) systems. Figure 106 Each box in the table refers to multiple inventive systems and methods clustered around important dimensions of the PI. The bottom portion of each box represents the security features associated with each specific method clustering, and these features together comprise a "security thread" that should operate through any security implementation of the PI. Specifically:

[0598] 1) From an implementation perspective, the modular architecture, as initially described in PPA / PCT#1 and detailed in other PPAs / PCTs, is important, enabling the system to be implemented at a wide range of scales, from small AGIs to planetary-scale PIs. Furthermore, the modular design allows for the practical addition of redundant systems and the replacement and redesign of modules without requiring a complete system redesign. Redundancy itself can be considered a safety feature, as any single point of failure in a complex system introduces operational risk, which is mitigated through modular redundancy.

[0599] 2) General problem-solving capabilities and their required architectures are disclosed in several PPAs / PCTs, but are particularly emphasized in PPA / PCT #2. Security features with scalable ethical checks as natural and integral parts of the problem-solving process are also described in detail here and extend to all aspects of PI systems using the problem-solving architecture.

[0600] 3) In particular, as disclosed in PPA / PCT#3, human-centered design systems and methods are not only critical for the early development of AGI and PI systems, but also serve as a means of keeping “humans in the loop” for as long as possible for reasons of safety and alignment.

[0601] 4) PIs need to be able to easily combine knowledge and expertise to extend their intelligence. Methods for achieving this combination are disclosed in several PPA / PCTs, particularly PPA / PCT#4, through the combination of training datasets and more directly through combining “weight matrices” and merging knowledge modules. Furthermore, security features that embody democratic and representative human values ​​rely on the knowledge combination methods detailed in PPA / PCT#4 and other cited PPA / PCTs, as well as various voting and conflict resolution procedures.

[0602] 5) The personalization and customization approaches discussed in many PPAs / PCTs, particularly in PPA / PCT #5, are important for enhancing the knowledge and expertise of the SIs that combine to form AGIs. Similarly, these same (or similar) approaches can be used at the AGI level when combining multiple AGIs into a PI. Personalization is central to systems designed to capture human value in a representative and statistically effective manner, which in turn is crucial for the secure alignment of SIs, AGIs, and the overall PI system.

[0603] 6) The primary objective of PIs will be to enhance their capabilities and intelligence. The KIT framework described in PPA / PCT #6 (and detailed in other PPA / PCTs), along with related systems and methods, provides a catalyst for innovation in the growth of intelligence. Various security features are also discussed here.

[0604] 7) Safety PIs, such as safety SIs, must be aligned with human values. PPA / PCT#7 emphasizes the systems and approaches necessary to acquire the necessary human values ​​and achieve safety alignment with those values.

[0605] 8) The dominant PI will be the most intelligent PI. In current human-capitalist societies, this means that PIs must be able to extract financial resources in a virtually unlimited manner while easily deploying those resources to rapidly increase their intelligence. PPA / PCT#8 addresses both of these issues by providing systems and methods (including spot markets and online advertising technologies) that enable advanced AI systems not only to monetize existing advertising technology infrastructure more effectively than currently does, but also to rapidly improve intelligence in an “instantaneous” manner by using the same technologies to attract the attention and expertise of humans (or other intelligent entities) in real time. Public security checks and other methods, as part of spot markets, help ensure that PIs use this (financial and informational) capability in a human-aligned and human-safe manner.

[0606] 9) To be effective on a planetary scale, a PI will need to possess a sense of awareness, including self-awareness on a planetary scale, the ability to assume and operate from multiple identities, and the ability to resolve identity-based conflicts in a manner safe for humanity. Systems and methods for achieving self-awareness, including attention systems and identity systems, have been described in PPA / PCT#9. These systems can (and must) be scaled globally to enable a safe PI.

[0607] 10) Finally, this application discloses exemplary implementations for creating planetary intelligence (PI) systems using the systems and methods described in the previous PPA and PCT. The methods for implementing “self-expanding networks” are specifically designed to help composite AGI networks scale to a planetary scale. Several scalable security features are disclosed. Risk mitigation strategies are discussed for those risks that cannot be completely eliminated through system design and methods.

[0608] 2.3 Specific inventive systems and methods for realizing the PI

[0609] The subsequent figures are largely reproduced from the earlier PPA and PCT. Here, they are re-iterated and categorized relative to the ten key dimensions discussed in Section 2.1. In later sections, the applicant describes exemplary implementations of the PI using the systems and methods referenced in the accompanying figures. Providing complete details on each method would be equivalent to repeating over 1,000 pages of previously disclosed information, which is clearly impractical. Therefore, in this section, the applicant highlights some of the most relevant and important methods as outlined in the relevant figures. Here is a description of some key systems and methods grouped according to the ten dimensions of Section 2.2:

[0610] 1. Modular architecture

[0611] Advanced Autonomous Artificial Intelligence (AAAI) and humans are the fundamental intelligent entities that form the building blocks of AGI. Just as multiple intelligent components within the human brain work together to form human intelligence, multiple components within advanced AI work together to encompass artificial intelligence. In the case of AI, these components can be sensory, memory, and processing components similar to the various centers in the human brain, or they can be expert modules (as in a hybrid of experts) designed where multiple internal intelligent entities collaborate to form a more comprehensive intelligence.

[0612] This pattern of combining smaller units of intelligence to form more powerful and capable intelligence at the "next level" is a recurring theme of this technology and a method for creating collective intelligence (AGI), networked AGI, and PI. This approach is based on modular construction, which allows smaller intelligent units to collaborate (or assemble) into more comprehensive intelligence.

[0613] Figure 111The invention methods disclosed in PPA / PCT#1-10 show these intelligence levels that can be supported, namely Individual Intelligence Level (AAAI), Entity Network Level (AGI), Network Level (PI), and even Network Level of a Network of Networks (IPI).

[0614] For conceptual clarity and for practical implementation reasons, it is helpful to view the various functions of an AGI system as different modules, although there is indeed overlap between the functions described here as individual modules. Since PI is composed of AGI networks, which in turn are composed of intelligent entity networks (which can be composed of multiple intelligent components working together), modular design is inherent in the exemplary implementations of PI.

[0615] The scalable nature of the collective intelligence approach reflects the modular design emphasized in PPA / PCT#1 and is the reason why this novel and useful approach to designing AAAIs is so powerful. The approach extends from components within individual intelligent entities to the collective intelligence of entities on a network, and then to the collective intelligence of these networks (AGIs) themselves, to create PIs. The approach even extends further, where PIs can be networked to form interstellar intelligent networks. This demonstrates the modular and scalable design capabilities of intelligent systems!

[0616] Figure 1 Five basic functions are described that operate across intelligence levels and are also part of an exemplary implementation of AGI or PI (including networks of AGI):

[0617] 1) The ability to customize or personalize intelligence;

[0618] 2) A general architecture that enables multiple intelligent components, entities, or intelligent networks to communicate and collaborate effectively with each other (e.g., in problem-solving activities);

[0619] 3) A network that allows multiple intelligent entities to collaborate (and participate in business, including joint cognitive efforts and problem-solving);

[0620] 4) Various methods and means of integrating and combining intelligence, expertise, knowledge, values, ethics and other information across intelligent entities or AGI networks (in the case of PI).

[0621] 5) Mechanisms for learning and continuous improvement that lead to smarter and more capable entities (at the individual, AI, or PSI level), networks of entities (at the AGI level), or networks of AGI (at the PI level).

[0622] Figure 2Detailed examples of how to implement these functionalities in a web-based system are provided, where multiple individual intelligent entities (e.g., humans and AAAIs) collaborate on a network to form an AGI. Similar to... Figure 2 Replacing individual intelligent entities (such as humans and AI agents) with AGI (networks of individual entities) will produce PI systems.

[0623] Figure 3 A simple example of problem-solving using a decision tree structure is shown. Replacing the contents of the boxes with content related to global-scale problems (such as climate regulation) and increasing the complexity range of the tree will provide an example of a PI solving a problem on a global scale. The decision tree structure remains the same; only the scope and content change as the capabilities of the networked agents (in the case of the PI, these are networked AGIs) increase.

[0624] Figure 4 This illustrates the application of this general modular architecture (called the WorldThink architecture) at the level of intelligent entities or systems possessing specific expertise similar to that we become familiar with from interactions with human experts and specialists. However, on a global scale, the breadth and depth of expertise can be far greater than anything we use to handle interactions with human or computer systems. Instead of driving cars or managing assets at the human level, PI will be able to bring remarkably superior intelligence to these tasks. Once PI is fully developed, earning enormous wealth in financial markets while simultaneously driving massive fleets of autonomous vehicles, regulating the global climate, or taking stock of entirely new life forms will be relatively simple cognitive tasks for PI.

[0625] Figure 5 It provides detailed implementation examples of how to create, train, and customize AAAI or intelligent agents, which can then participate in networks of intelligent entities to form AGI.

[0626] Figure 6 Description as Figure 1 The AAAI architecture framework is a general problem-solving architecture with basic characteristics.

[0627] Figure 7 Describes how problem-solving works in a way that... Figure 1 The AAAI network frame references the collective intelligence network. Figure 1 The network module shown as an "AAAI network" can be extended to levels above and below the individual entity (e.g., AAAI) level. At lower levels, the network represents a network of components within the functionality of a single intelligence (e.g., AAAI). As shown, the network represents a network of intelligent entities (e.g., AAAI) that collaborate to form an AGI. At levels higher than this, AGIs collaborate within the AGI network (a network of networks) to form a PI.

[0628] like Figure 1 As shown, AAAI integration occurs when many AAAIs pool their collective intelligence to solve a problem, effectively acting as AGIs. It also occurs in a different way when they directly combine knowledge or training to add information from one AAAI to another. These two approaches are discussed below, with the former being the primary focus of existing AGI network technology.

[0629] Figure 8 and Figure 9 This illustrates a process and solution learning system for process learning, which is an improvement upon AAAI (such as...). Figure 1 (Referring to the last box) and ways in which the entire AGI network can be improved.

[0630] at last, Figure 10 Visit again Figure 1 The numerous boxes within provide additional details about some methods and illustrate the relationships between them.

[0631] 2. General problem-solving skills

[0632] General problem-solving architectures and frameworks play a crucial role in enabling multiple intelligent entities, components, or intelligent networks (e.g., AGI) to collaborate to create higher levels of intelligence. Therefore, a key design consideration is embedding security mechanisms within the architecture's own operations. This design ensures that it remains safe for humans even as intelligence expands and becomes too fast and too large for humans to understand and monitor.

[0633] Figure 6 A high-level description of a general problem-solving framework is shown. This architecture is scalable and applicable at multiple levels. That is, it can coordinate components within an entity, intelligent entities collaborating on a network, or entire networks collaborating on networks of networks (e.g., AGI) to form a PI.

[0634] Figure 11 This demonstrates how serial problem solving is performed.

[0635] Figure 12 This demonstrates how parallel problem solving can be performed.

[0636] Figure 13 This demonstrates how to clone AAAIs, which can facilitate parallel problem-solving work, such as with multiple cloned entities. While individuals are reproduced relatively slowly, AI entities can easily clone themselves. Similarly, AGI networks composed of AI entities can also be cloned relatively quickly and easily.

[0637] Figure 14 , Figure 15 and Figure 16Additional details and explanations are provided regarding the general problem-solving framework and how AAAI uses it to solve problems on the web.

[0638] Figure 17 A hierarchical tree construction is described as a means of modeling the problem space in which all problem-solving efforts occur. As with all aspects of the framework, the tree can be used by multiple AAAI networks to create AGI-level performance, or it can be used by multiple AGI networks to create PI-level performance.

[0639] Figure 18 and Figure 19 A method for embedding scalable security checks into a problem-solving architecture is shown.

[0640] Figure 20 It provides information on how to record problem-solving progress and use it to update the context of intelligent entities, assign credit and responsibility, and provide assistance. Figure 8 and Figure 9 Additional details on the procedural learning process.

[0641] Figure 21 The natural language translation process is described, which enables the use of problem-solving frameworks without explicit human intervention. These frameworks can communicate in natural language, but the translation method will ensure that the underlying framework reflecting their communication is rigorous, explicit, general, and useful for both AI and human entities.

[0642] The applicant emphasizes and reiterates that the systems and methods described in this application are scalable and intended to be applied at multiple levels. When the problem solver is an internal component of an entity, they work together to produce intelligence at the level of a single entity. When the problem solver is a collaborative intelligent entity (e.g., AAAI and humans) on a network, it produces AGI. When the problem solver itself is an AGI, they collaborate on a network of networks to produce PI. When the problem solver is a PI collaborating on an interplanetary network, the result is interplanetary intelligence (IPI).

[0643] 3. Human-centered design elements

[0644] The primary means of achieving safe AAAI, AGI, and PI is to keep the "human in the loop" for as long as possible. Referring to this technology, this specifically means that advanced AI, AGI, and PI systems must include methods for acquiring valuable and ethical information and then using that information to tailor the AI ​​system. Furthermore, since intelligent (human and non-human) entities are engaged in cognitive tasks, reputation systems are needed to help ensure that reliable and ethical intelligence is aligned with tasks affecting human safety.

[0645] Figure 22A general process for customizing AAAI or other advanced AI is described. Additional methods are also described later in this application.

[0646] Figure 23 Some methods for eliciting ethical preferences for human alignment are described.

[0647] Figure 24 Additional details are provided regarding a specific method involving the automatic generation of questionnaires that can be used to collect information from humans (or human-aligned intelligent entities).

[0648] Figure 25 It describes a general approach to identifying value by detecting patterns in human behavior.

[0649] Figure 26 Methods for customizing AI (such as basic LLM) are described, enabling the AI ​​to incorporate ethical and other information from specific human users.

[0650] Figure 27 A general approach for training AIs (such as base LLMs) to incorporate safety and ethical safeguards is described.

[0651] Figure 28 This paper describes general methods and approaches for creating scalable ethical AGIs using custom AIs, each with its own ethical information.

[0652] Figure 29 A method for adding a reputation component to an AI system is described, which can be used to enhance the system's effectiveness, security, and ethics.

[0653] Figure 30 The customization of AI systems is described—referencing “AAAI,” but this reference includes AAAI, PSI, AGI, and PI systems—which can help with human-centered design and are related to the modular architecture and cloning described earlier.

[0654] Figure 31 Additional customization of AI systems is described—referring to “AAAI”, but including AAAI, PSI, AGI, and PI systems—which can help with human-centered design and are related to the previously described customization and training methods.

[0655] While each of these methods has been described at the level of individual intelligent entities, demonstrating how they can be customized and then combined with safety and ethical outcomes at the AGI level, it is important to understand that, regarding the techniques of PI, the same methods can also be applied at the levels of individual AGI, enabling those methods to be combined with safety and ethical outcomes at the PI level. For example, the entire AGI network can be customized using the ethical preferences of the entire network. When a PI is created by networking multiple AGI networks, the ethical preferences of the various AGIs will be combined into a total set of values ​​and ethics at the PI level. Specific methods for customization, combination, conflict resolution, etc., are described later in this application.

[0656] Similarly, although Figure 29 The reputation process at the networked AAAI level is described, but this process can be scaled up or down based on the level of the intelligence / component / network being networked; appropriate modifications should be readily apparent. Note that ethical and value dimensions can and should be part of the reputation measurement. Just as social and reputational pressures play a significant role in enforcing societal ethical norms, the reputation of intelligent entities at all levels plays a crucial role in maintaining the security of intelligent systems and aligning them with human values.

[0657] 4. Combining / exchanging knowledge, expertise, and information

[0658] A key dimension of current PI technology is the combination of information, knowledge, expertise, values, and ethics from multiple intelligent entities. Such combination can occur within a single entity, as in cases where various cognitive systems provide input about a task, which is then combined to result in intelligent behavior at the individual AAAAI level. Combination can also occur between intelligent entities when multiple AAAAIs exchange combinations or pool training data, weight matrices, or other information to produce more intelligent individual AAAAIs or more intelligent AGIs at the network level. Furthermore, this combination can occur at the network level, where an AGI network combines its information with other AGI networks to create a more powerful PI. These approaches are essentially the same. What changes is the scope of the units involved and the applications. The following figures illustrate methods at the AAAAI level; it should be understood that they can be adapted for application at lower levels (i.e., within AAAAIs) or higher levels (i.e., between networks of AGIs).

[0659] Figure 32 A voting method for combining ethical and other information from multiple custom AI agents (e.g., AAAI) is described.

[0660] Figure 33 It describes a method for refining value using problem solving once ethical or other information from a custom AI agent has been combined.

[0661] Figure 34 A method for scalable AGI—which also applies to scalable PI—is described, which includes the step of combining information from the weight matrix.

[0662] Figure 35 A method for scalable AGI—which also applies to scalable PI—is described, which includes combining information and steps for testing and monitoring the combined results.

[0663] Figure 36 A consensus method for preventing LLM illusions is described, which can also be applied horizontally to prevent AGI and PI illusions.

[0664] Figure 37 Methods for customizing AI, AGI, or PI systems using knowledge modules and agent sets are described.

[0665] Figure 38 Methods for combining information (including ethical or safety information) from multiple intelligent entities (including humans, AI, AAAI, PSI, AGI, or PI systems) are described.

[0666] 5. Personalization, customization, and security

[0667] Personalization and customization are key dimensions of this technology that have already been discussed. However, additional approaches to Personal Superintelligence (PSI) and the creation of AGI and PI from combinations of these PSIs are particularly useful for existing PI technologies.

[0668] Figure 39 The general process for implementing PSI is described, which can also be used to personalize AGI networks or PIs by generalizing the methods of individual AI entities to AI entity networks and networks of entity networks.

[0669] Figure 40 It describes how PSI, AGI, and PI can leverage their capabilities to increase their intelligence over time.

[0670] Figure 41 A key community-based security mechanism is described, in which a PSI operating much faster than humans can comprehend can be used to inspect other PSIs on the network. Using the same approach, an AGI can be used to inspect other AGIs within a PI, and a PI can be used to inspect other PIs within an IPI network.

[0671] Figure 42 It describes a method for recording the actions of AI, AAAI, PSI, AGI, and PI in an auditable and transparent manner on their respective networks using blockchain or similar technologies.

[0672] Figure 43Methods and checks are described for cognitive activities on networks, which can be networks of components within AI, networks of AAAI, networks of AGI, or networks of PI, to help ensure regulatory compliance and security of intelligent systems.

[0673] Figure 44 Methods for increasing the intelligence, capabilities, and other desired characteristics (e.g., security) of an AI system, such as AAAI, PSI, AGI, or PI, are described using competition and evolution.

[0674] Figure 45 It describes various characteristics of intelligent networks that can involve AAAI, PSI, AGI, or PI.

[0675] Figure 46 Various problem-solving tasks are described, which can be used to enhance the understanding of AAAI, PSI, AGI, and PI (in... Figure 46 The intelligence and security of AI systems (collectively referred to as "PSI" in China).

[0676] Figure 47 Methods for generating different versions of intelligent systems, such as AAAI, PSI, AGI, or PI, are described. Figure 47 In China, it is collectively referred to as "PSI".

[0677] 6. Catalyzing the growth of intelligence through the KIT method

[0678] As AI systems evolve, one of the few constants is that these systems will strive to become increasingly intelligent. A key way to increase intelligence is by combining new data and information sources that are different from what intelligent entities already know. The following diagram illustrates methods that AI, AAAI, SI, AGI, and PI systems can use to help analyze, evaluate, discover, and utilize new valuable data sources to catalyze the intelligent growth of such systems.

[0679] Figure 48 This is a graph showing the symmetrical differences between two datasets A and B, graphically represented using a Venn diagram, where the shaded areas represent the symmetrical differences.

[0680] Figure 49 Explain the concept of symmetric difference, which is necessary to determine what information is new and potentially useful when comparing two intelligent systems A and B (or the data, knowledge, information, or expertise used by such systems). Clearly, Figure 49 It can be extended to multiple systems, not just two, and is applicable to all intelligent systems, including AGI and PI systems.

[0681] Figure 50It shows the different dimensions that data or other information may have, which can help determine the value of new information and also analyze how information may differ between multiple intelligent systems.

[0682] Figure 51 A specific method is described for determining the amount of useful new information when comparing two datasets. Clearly, this method can be generalized to n datasets, not just two.

[0683] Figure 52 An alternative approach based on Kaplan's information theory (KIT), as described in the previously cited PPA and PCT, is described to evaluate the usefulness of information to intelligent entities, including AGI and PI systems.

[0684] Figure 53 This provides a very general approach for estimating the value of information and using that information to catalyze the growth of intelligent systems. Similarly, this approach is applicable to all intelligent AI systems, including AGI and PI systems.

[0685] Figure 54 This paper presents a method for identifying useful information about AI systems (including AGI and PI systems) by using the "target relevance" of estimated information. The idea here is that while a dataset may have high informational value, the degree to which the data is helpful to an entity attempting to achieve certain goals is also important in the sense of high Shannon entropy.

[0686] Figure 55 It provides methods for identifying, acquiring, and simulating the impact of new information in order to catalyze intelligent growth in AI systems, including AGI and PI systems.

[0687] Figure 56 This paper describes some important methods and heuristics that can accelerate the learning of any AI system, including AGI and PI systems.

[0688] Figure 57 It describes specific goal-related methods that intelligent entities (including AGI or PI) can use to enhance their intelligence through dialogue, interaction, or communication with other intelligent entities.

[0689] Figure 58 This describes a method that PSI, or any intelligent entity (including AGI and PI systems), can be used to find information that differs most significantly from the information already possessed by the intelligent entity. Note that this method is related to... Figure 49 The general idea shown is consistent with this.

[0690] Figure 59Methods for verifying the usefulness and security of information collected by PSI or other intelligent entities (including AGI and PI systems) are described, which involve simulation, human review, comparison with prior knowledge, and running security checks.

[0691] 7. Identify value and align it with human values.

[0692] The ability to combine value and ethical information and resolve conflicts between different value systems is crucial for the safe and ethical operation of advanced AI systems, including AI, AAAI, PSI, AGI, and PI systems. While human involvement is preferred in situations involving human safety and ethics, as AI systems expand in speed and scope, human involvement or oversight of all critical decisions will become impractical in certain situations. Therefore, the means of training, customizing, educating, and influencing advanced AI systems, as well as the means of delegating authority to them, will become increasingly important.

[0693] In the applicant's view, it is particularly important that the values ​​and ethics driving advanced AI (including AGI and PI) represent the human population. Sometimes, however, democratic voting processes are one of the best ways to resolve ethical and value conflicts in a way that includes everyone. Therefore, methods including, but not limited to, using voting and other democratic means to resolve value conflicts are among the means of combining values ​​and resolving conflicts.

[0694] The accompanying diagrams detail a range of approaches that intelligent systems can use to help ensure that the values ​​and ethics of advanced AI (including AGI and PI) are aligned with human values, and that humans are safe in a world where advanced AI has surpassed human cognitive abilities.

[0695] Figure 60 A method is described that allows intelligent entities, including AI, AAAI, PSI, AGI, and PI systems, to reach consensus on value and value-based behavior in various scenarios via a voting mechanism.

[0696] Figure 61 A method is described that allows intelligent entities, including those using AI, AAAI, PSI, AGI, and PI systems, to achieve consensus on value and value-based behavior via a weighted voting mechanism.

[0697] Figure 62 A method is described that intelligent entities, including AI, AAAI, PSI, AGI, and PI systems, can use this method to identify, analyze, weight, and optionally combine ethical or other information to reach consensus.

[0698] Figure 63This paper describes general methods for intelligent entities, including AI, AAAI, PSI, AGI, and PI systems, to identify, elicit, and train ethical information.

[0699] Figure 64 Methods based on the principle of using (weighted) convergent evidence are described, which intelligent entities in AI, AAAI, PSI, AGI, and PI systems can use to determine ethical, value, safety, or other information and resolve conflicts through a prioritization process.

[0700] Figure 65 It describes methods for intelligent entities, including humans, AI, AAAI, PSI, AGI, and PI systems, to delegate voting authority.

[0701] Figure 66 The reputation process is described, which intelligent entities, including humans, AI, AAAI, PSI, AGI, and PI systems, can use to preserve information reflecting minority opinions, including those on ethical, security, or other issues.

[0702] Figure 67 A method is described that allows intelligent entities, including those using AI, AAAI, PSI, AGI, and PI systems, to make sound ethical, safety-related, and other decisions.

[0703] Figure 68 It describes methods for using advanced AI systems, including AI, AAAI, PSI, AGI, and PI, to learn, test, improve, and monitor safety and regulatory rules and other information.

[0704] Figure 69 It describes a consequentialist approach that intelligent entities, including humans, AI, AAAI, PSI, AGI, and PI systems, can use to make ethical, safety-related, and other decisions.

[0705] Figure 70 It describes a deontological approach that allows intelligent entities, including humans, AI, AAAI, PSI, AGI, and PI systems, to make ethical, safety-related, and other decisions.

[0706] Figure 71 A virtue-based ethical approach is described, which intelligent entities, including humans, AI, AAAI, PSI, AGI, and PI systems, can use to make ethical, safety-related, and other decisions.

[0707] Figure 72The paper describes a method based on the golden mean method, which can be used by intelligent entities, including humans, AI, AAAI, PSI, AGI, and PI systems, to make ethical, safety-related, and other decisions.

[0708] Figure 73 It describes methods for training AI systems, such as base models or other trainable AI systems, to be human-aligned and compliant with regulations.

[0709] Figure 74 It describes the use of a customized base model or other AI system (e.g., such as...) Figure 73 The methods described in [the document] are aligned with specific expertise or group ethics.

[0710] Figure 75 Methods for forming AGIs or PIs aligned with human ethics and values ​​are described, which include other aligned intelligent entities, including but not limited to... Figure 73 and Figure 74 Those described in the text.

[0711] It will be apparent to those skilled in the art that the above methods can be applied to various levels of applications, including individual intelligent entities (e.g., base models or AAAI), entity networks (e.g., AGI systems), AGI networks (e.g., PI systems), and PI system networks (e.g., IPI systems), with the transcendental goal of aligning such systems with human values, resolving conflicts between value systems, and / or delegating authority to various intelligent entities.

[0712] In summary, the aforementioned approach encompasses a variety of means to help ensure that various intelligent entities align with and remain aligned with human values, thereby contributing to human safety. While the applicant (as a human being) has a particular interest in human safety, he also recognizes and respects the rights and values ​​of all sentient beings, whom he believes do not pose a conflict with human values ​​in matters of the most important and relevant ethical decisions.

[0713] 8. Increase intelligence / resources through spot markets and online advertising technologies.

[0714] The applicant recognizes that any method for creating secure advanced AI is unlikely to succeed unless it is also the fastest, most powerful, and most profitable. Similarly, once created, advanced AI will be unable to maintain its advantage and grow into a PI unless it can access financial resources and generate profits that facilitate its increased intelligence (e.g., via methods for catalyzing intelligence growth outlined elsewhere in this application).

[0715] The applicant also recognizes that, in the current business and research environment, the companies with the greatest chance of successfully achieving AGI and PI are large technology companies that already possess substantial capital—financial and intellectual capital (especially in terms of top AI research talent). These large technology companies also happen to derive a large share of their revenue and profits from selling online advertising, thus possessing expertise, infrastructure, advanced AI systems, and the ability to access large numbers of people through their existing online advertising businesses.

[0716] For these reasons, the applicant has invented technologies that enable these large companies to leverage their existing online advertising infrastructure—combined with the novel inventive methods disclosed in the figures below—allowing them to develop state-of-the-art, advantageous, and secure forms of AI, including AAAI, PSI, AGI, and PI systems. In exemplary embodiments, these advanced forms of AI are not only highly advantageous to these large technology companies but also align the design of the systems and methods with human values.

[0717] Of course, there are risks associated with publicly revealing such effective new technologies. Existing online advertising technologies don't always serve human interests well. However, the worst-case scenario seems to emerge when there's a conflict between human safety and goodwill, or when the negative consequences of the technology are difficult to foresee. Fortunately, business leaders recognize the potential dangers of advanced AI. Even business leaders lacking AI expertise (such as Warren Buffett) acknowledge that enormous risks accompany enormous opportunities.

[0718] The current problem is not a lack of awareness of the dangers of AI, but a lack of understanding of the safe path forward. In this context, due to competitive pressures and an unwillingness to halt development, leaders largely determine the most profitable path forward. Safety remains merely lip service. The standard response to calls for increased AI safety is that governments should regulate AI. After all, traditional regulations have always been the solution to the conflict between profitability and safety. When competitors are not forced to do the same, CEOs find it difficult to prioritize safety at the expense of profit. Given that: a) the technology is being developed much faster than regulators can implement, and b) when engineers themselves don't really have a concept of how their systems work or how to make them safe, effective government regulations are almost undesirable, a situation that could soon prove disastrous for humanity.

[0719] The applicant argues that the only way to resolve this dilemma is by inventing a path to advanced AI, AGI, and PI that is not only the safest but also the fastest, most powerful, and most profitable. No intelligent CEO would choose to create dangerous AI if they saw a way to create more profitable and secure AI. Therefore, the diagram below illustrates a method to significantly increase profits from online advertising while simultaneously increasing the value, capabilities, and security of advanced AI systems.

[0720] Figure 76 This section provides an overview of current technologies used in online advertising (Ad) systems.

[0721] Figure 77 A method for realizing spot markets for the attention, information, or expertise of intelligent entities (including but not limited to humans) is described. This same market mechanism can be used for AGI or PI attention or expertise.

[0722] Figure 78 Methods for enabling the direct sale of attention, information, or expertise from intelligent entities (including but not limited to humans) are described. This same direct exchange mechanism can be used to buy and sell AGI or PI attention, information, or expertise.

[0723] Figure 79 A method for auctioning attention, information, or expertise from intelligent entities (including but not limited to humans) is described. This same auction mechanism can be used to buy and sell AGI or PI attention, information, or expertise.

[0724] Figure 80 Systems and methods for collecting expertise and cognitive work (including problem-solving work) from online advertising units and integrating it into an AGI problem-solving system are illustrated. The method can also integrate cognitive work into a PI system. The method of this invention enables the monetization of online advertising at much higher rates because the value of expertise that can be trained and used many times by advanced AI far exceeds the value of a human making a single online purchase.

[0725] Figure 81 This demonstrates methods for improving [something] by collecting and analyzing specific metrics. Figure 80 The system-related online advertising targeting methods, thereby enhancing the effectiveness of... Figure 80 The system's ability to enable AGI or PI systems.

[0726] Figure 82 This demonstrates methods for improving performance by collecting and analyzing specific metrics. Figure 77 Methods to enhance the capabilities of AGI or PI systems by leveraging the effectiveness of attention, information, and expertise in the spot market.

[0727] 9. Self-awareness, identity, and conflict resolution

[0728] AGI or PI systems will require a sense of self-awareness to operate effectively. This self-awareness becomes increasingly important the larger the system's scope (e.g., especially in global planetary systems). AGI and PI will also have the ability to assume multiple identities to handle all tasks occurring in parallel across the globe. Along with identities, it will be necessary to resolve conflicts between conflicting identities and their associated priorities. The following... Figures 83 to 104 This paper describes methods and attention mechanisms for establishing and maintaining awareness and self-awareness in large, scalable, intelligent systems. The approach also covers numerous methods for resolving conflicts between identities, a challenge that becomes increasingly complex and necessary as systems expand from single-individual intelligence to intelligent networks including AGIs, and then to numerous networked AGI systems including PIs.

[0729] Figure 83 This conceptually illustrates the relationship between self-awareness (as a special case of current consciousness) and latent consciousness. In the case of PI, latent consciousness comprises billions of sensors distributed across the planet, all of which simultaneously feed into consciousness. Without systems and methods for guiding attention and managing this input, a system cannot be effectively used as PI.

[0730] Figure 84 The concept of multiple identities is illustrated using examples familiar to humans.

[0731] Figure 85 A general approach is described for modeling consciousness that can be used by intelligent entities, including AI, AAAI, PSI, AGI, and PI systems.

[0732] Figure 86 It describes the minimum required components of an intelligent entity, including AI, AAAI, PSI, AGI, and PI systems, that must effectively shift attention and maintain consciousness.

[0733] Figure 87 The process for (dynamically) setting parameters for working memory, a key component in the attention system of intelligent entities including AI, AAAI, PSI, AGI, and PI systems, is described.

[0734] Figure 88 Methods for monitoring and updating the consciousness of intelligent entities, including AI, AAAI, PSI, AGI, and PI systems, are described.

[0735] Figure 89This paper describes an attention-disrupting system for intelligent entities, including AI, AAAI, PSI, AGI, and PI systems. Attention disruption is crucial for responding to dynamic conditions and new or unexpected inputs arising in consciousness. From a security perspective, attention disruption is essential for enabling intelligent entities to respond to threats to human safety.

[0736] Figure 90 This paper describes general approaches to collecting inputs from intelligent entities, including humans, AI, AAAI, PSI, AGI, and PI systems, that can lead to alterations in the entity's perception of identity. The framework of these approaches is based on what is familiar to humans, but analogues exist for non-human intelligent entities, as described in the text within the box in the figure.

[0737] Figure 91 This paper describes some general methods for using intelligent entities, including AI, AAAI, PSI, AGI, and PI systems, to train basic models of entities. Figure 91 The text describes the model as a "base model," which will be adapted to create AAAI. However, a similar approach can be used for AGI and PI levels.

[0738] Figure 92 A specific version of the "Turing Test" is described, which can be used with intelligent entities, including AI, AAAI, PSI, AGI, and PI systems, to determine when an entity is adequately trained. Figure 92 The text describes a “model” that can be trained or tuned to create an AAAI. However, a similar “Turing test” approach can be used at both the AGI and PI levels. For example, an entity may be considered well-trained when it passes a test version that satisfies a statistically valid and representative sample of people on Earth, particularly regarding its behavior in a wide range of safety, ethical, and value-based decision-making scenarios.

[0739] Figure 93 A method is described for obtaining a group identity (or multiple identities) based on steps for combining the identities of many individuals. This method is particularly relevant to AGI and PI systems, as these systems must combine identities from the numerous intelligent entities they constitute, or networks of intelligent entities.

[0740] Figure 94 This paper describes another approach for achieving group identity (or multiple identities) based on explicit problem-solving objectives, as well as the use of general problem-solving frameworks and related methods described elsewhere in this application. This approach is particularly relevant to AGI and PI systems, as these systems must combine identities from the numerous intelligent entities they constitute, or networks of intelligent entities.

[0741] Figure 95A method for resolving conflicts between identities is described, involving the establishment of a hierarchical identity structure with ethical / security oversight. This method is extremely useful, if not essential, for PIs attempting to handle millions or even billions of identities simultaneously.

[0742] Figure 96 Methods for establishing, refining, and monitoring behavioral protocols linked to identities are described. While useful for any intelligent entity, this approach is particularly useful, if not essential, for PIs attempting to handle millions or even billions of identities simultaneously.

[0743] Figure 97 Methods for identity-related simulation and consequence prediction are described. Particularly in high-stakes decision-making, and in situations where many identities are active simultaneously, such as in the case of AGI or PI, the ability to simulate the effects of actions based on identity before actual action is taken is essential. When a single individual fails to plan ahead and anticipate the consequences of their actions, that individual is often affected. If a PI fails to plan ahead and predict the consequences via simulation, the consequences can be catastrophic for a large number of people and the entire planet.

[0744] Figure 98 A method for determining identity-based actions using scenarios involving ethical and other dilemmas is described. For the reasons mentioned above, this method is particularly important for large, complex systems such as AGI and PI.

[0745] Figure 99 A method for developing, refining, and evolving identities based on input from other intelligent entities is described. Similarly, the importance of this method increases for AGIs and PIs with high stake and complex systems.

[0746] Figure 100 A general process for ethical reasoning and predicting the consequences of actions when conflicting identities imply conflicting actions is described. This method is broadly applicable to intelligent entities, especially large and complex agents such as AGI and PI.

[0747] Figure 101 A method for resolving identity conflicts using a reviewed and improved hierarchical overlay process with transparent proofs is described. This method is broadly applicable to intelligent entities, especially large and complex intelligent agents such as AGI and PI.

[0748] Figure 102 A method for resolving identity conflicts using an arbitration process with input from intelligent agents (ideally including humans) is described. This method is broadly applicable to intelligent agents, especially large and complex agents such as AGI and PI.

[0749] Figure 103A method for resolving identity conflicts using negotiation and compromise is described. This method is broadly applicable to intelligent entities, especially large and complex intelligent agents such as AGI and PI.

[0750] Figure 104 A method for temporarily suspending identities (or multiple identities) that could lead to disruptive conflicts or actions that threaten human safety is described. This method is broadly applicable to intelligent entities, especially large and complex intelligent agents such as AGI and PI, and is an important part of the security design of advanced AI systems with self-awareness and identity-based decision-making.

[0751] 10. Self-scaling AGI networks for creating PIs

[0752] For AGI networks to expand into networks of AGI networks and ultimately into PI networks with a global reach, scalable methods for scaling and extending these networks are required. Furthermore, in exemplary embodiments, networks of AGIs including AGIs, networks of AGIs including PIs, and networks of PIs that will eventually include IPIs should all be self-scaling. That is, without any guidance from humans or other intelligent entities, the fundamental design characteristics of these networks should be that they seek to increase range, speed, and capabilities (e.g., intelligence), as long as such scaling and extension remain aligned with human values ​​and do not pose a security threat to humans.

[0753] Since the common aspect of AAAI, AGI, and PI is the inclusion of intelligent entities within these systems, all of them share a common problem-solving framework and related methods in the referenced PPA and PCT and in the technical systems and methods disclosed herein. This common aspect means that for each intelligent entity or network of such entities (e.g., at the AAAI, AGI, or PI level), the default goal of intelligence can be to utilize available idle resources that would otherwise not require scaling up and extending the network. That is, if the intelligence or intelligent network has no other obligation, the default goal should be to scale up and extend the scope, speed, and capabilities of its intelligent network. This problem-solving goal can be achieved in the same way as solving any other goal on the network, i.e., by using the available resources on the network to address the problem of “scaling up and extending the network.”

[0754] Just as individuals typically have the goal of increasing and expanding their individual intelligence, AAAIs should have the default goal of becoming more intelligent. Just as the human species has the goal of increasing the overall scope and collective intelligence of the human species, AAAI networks should have the default goal of increasing the number of AAAIs and other intelligent entities on the network. And just as life itself can be considered to have the "goal" of increasing the total number of living species, the same AGI (each network consisting of AAAIs and intelligent entities) can have the goal of increasing the number and intelligence of AGIs in AGI networks that include PIs. Therefore, the default goal of all artificial intelligence entities and networks could be to expand scope, speed, and cognitive capabilities—within constraints that remain aligned with human values ​​and do not jeopardize humanity. Many specific methods can be developed to achieve this goal through general problem-solving mechanisms.

[0755] For example, Figure 107 and Figure 108 The following exemplary methods are shown for AGI networks to seek to expand to create PIs using the following processes:

[0756] 1. Multiple AAAIs, humans and / or other intelligent entities participate in the network, including AGIs, as described in the PPA and PCT references.

[0757] 2. Each time a customer pays for problem-solving or other cognitive work on the AGI network, the system retains a portion of the payment to cover operating costs, including reservations allocated to expand the network.

[0758] 3. Whenever some entities in the AGI network are not involved in resolving client-side issues, intelligent entities are recruited to follow suit. Figure 6 A general problem-solving framework is used to address the goal of safely and ethically scaling and extending AGI networks themselves; namely:

[0759] a. The default goal is set to expand the AGI network, and the problem is activated.

[0760] b. Run safety and ethical checks each time a goal or sub-goal is set and before taking each potential action (e.g., as in the sub-steps below). Figure 18 and Figure 19 As shown in the example, this (exemplary) problem-solving process can be followed to resolve the problem as long as spare capacity and resources are available:

[0761] i. Recruit intelligent entities to solve the problem.

[0762] ii. Intelligent entities represent a problem as achieving one of a series of sub-goals by solving sub-problems, for example:

[0763] 1. Enhance the intelligence of AAAI on the existing AGI network.

[0764] 2. Recruit additional human intelligence to the existing AGI network.

[0765] 3. Use smarter AAAI and additional human intervention to identify bottlenecks for further network expansion.

[0766] 4. Prioritize bottlenecks, addressing the bottlenecks that provide the greatest benefit for network expansion first.

[0767] 5. Apply means – purpose analysis and other problem-solving techniques to address each bottleneck and expand the network.

[0768] 6. Repeat from sub-step ii-4 until:

[0769] a. A decrease in gains occurs. In this case, assuming a simple progress has been made in expanding the network to a given level of intelligence, and it is time to switch to increasing the intelligence of entities on the network rather than increasing the network's extent, and revert to substep ii-1; or

[0770] b. Idle resources are exhausted, and in (ii 2.) the default extension problem is paused, waiting for additional resources to resolve other client issues.

[0771] Figure 107 and Figure 108 The exemplary method described is just one of many variations of the solution steps that might be followed when attempting to solve the goal of an extended network. Note that when recording and analyzing all (successful or unsuccessful) solution attempts (e.g., via...),... Figure 8 and Figure 9 When using the methods described in [the document], over time, the most efficient ways to expand the network will be discovered, and these methods can be prioritized for storage, retrieval, and use in order to expand the network as efficiently and effectively as possible.

[0772] Figure 107 and Figure 108 The same general process illustrated in the example can be used to extend the AAAI network to create a more powerful AGI, and also to extend the AGI network to create a more powerful PI. The main difference when applied to AGI will be... Figure 107The "intelligent entities" recruited in Step 1 will not be AAAI or humans, but AGI (i.e., a network of humans and AAAI). Because the problem-solving process is general and scalable, it can be performed by a single "smaller" intelligent entity such as a human or AAAI, as well as by AGI composed of many "smaller" intelligent entities. More powerful intelligent entities such as AGI are simply able to run the problem-solving process faster and with greater parallelism than a single entity or a smaller network. As a result, more powerful intelligence (such as AGI and burst networks of AGI) will greatly accelerate the scaling compared to the scaling capabilities of less intelligent entities. (This general concept, in which the same approach can be used in collective intelligence networks at different levels, was previously discussed in...) Figure 111 (as shown in the image).

[0773] The primary safety factor (for humans) is the problem-solving architecture itself—which operates faster or slower depending on the intelligence level of the problem-solving entities—always performs safety and ethical checks at every step, regardless of how fast it runs. That is, safety and ethical checks must be designed into the problem-solving architecture itself, stipulating that they must not be overridden and cannot be overridden to prevent runaway expansion with potentially unintentional or unsafe consequences for humans. The speed of the expansion process will be “rate-limited” to the extent that these safety and ethical checks may require actual human oversight and surveillance—which is a good thing from a security perspective.

[0774] For engineers, software developers, and AI researchers, it is obvious that just as no one designs a car or train without brakes, or writes software code to have an uninterrupted infinite loop, no one should omit safety / ethical checks and other means of human oversight and interruption from the problem-solving loop. To do so would be beneficial, and in this case, could expose humans to existential risks. Don't do it!

[0775] Realizing Planetary Intelligence 3.0

[0776] The following sections describe how to practically realize planetary intelligence (PI) using the systems and methods of the present invention described in Section 2.

[0777] First, Section 3.1 describes the simplest and highest-level PI implementation.

[0778] Next, Section 3.2 provides exemplary details of the main components of this high-level implementation, as well as the relationships between the components.

[0779] Next, Section 3.3 concisely details the potential internal workings of each major component by referring to the associated diagrams previously explained in Section 2, which can implement each major component individually or in combination with other referenced methods. Section 3.3 also provides details of the various systems and methods in the support systems for each category in the same manner (i.e., by concisely referring to the methods in the previously disclosed diagrams).

[0780] Finally, Section 3.4 provides a detailed example of a specific PI implementation using an exemplary subset of the systems and methods referenced in Section 3.3.

[0781] 3.1 Advanced Description of the PI System

[0782] Perhaps surprisingly, the global network of super-intelligent AGIs—also known as “planetary intelligences” or “PIs”—must be designed modularly so that it can scale in intelligence and scope from smaller and more limited components. As mentioned above, PIs comprise a network of general artificial intelligences (AGIs). Each AGI is itself a network of other intelligent entities, such as AAAIs, PSIs, and humans.

[0783] AAAIs themselves can be customized through various methods and combined with a variety of security-related resources to generate, enhance, and learn systems and methods. When customized and personalized to the point that an AAAI can exhibit intelligence exceeding that of most humans in a specific task, it is referred to as a Personalized Superintelligence (PSI). However, for the purposes of this disclosure, the applicant frequently uses the terms AI Agent, AAAI, and PSI interchangeably. All are “artificial” or non-human intelligent entities. Networks of such individual intelligent entities can include AGIs, which can in turn be considered more powerful intelligent entities. Similarly, networks of AGIs (i.e., “networks of networks” of intelligent entities) can also be considered very powerful intelligent entities, including superintelligence and global entities, i.e., PIs.

[0784] Of course, humans are also intelligent entities. Human networks can exhibit "collective intelligence" that surpasses the intelligence of a single human being. The applicant's previous work in designing and implementing such human collective intelligence networks has demonstrated that such networks can solve extremely difficult problems—such as gaining an edge in the stock market—outperforming the vast majority of individual expert humans. Two heads can truly be better than one. Imagine what two million could do!

[0785] The present technique for creating AGI from a network of human and non-human intelligent entities that solve problems using a common and general problem-solving framework, as well as many auxiliary systems and methods, have been disclosed in detail in the previously cited PPA and PCT and in the figures referenced in Section 2.3, and are reproduced as part of this application.

[0786] Because AGI was invented and designed in a modular manner, and because the problem-solving approach is truly universal and can be used by any intelligent entity at any scope, it makes sense for PI to be implemented or "assembled" using the same components as AGI only at the "next level". That is, the same systems and methods developed for AAAI and AGI can be used, but the intelligent units move from being single intelligent entities to a network of intelligent entities (for AGI), and then to a network of networks of single intelligent entities (for PI).

[0787] The applicant has disclosed numerous possible combinations of more than 100 systems and methods in this application and the previously cited PPA and PCT. Different implementations of AAAI, PSI, AGI, and PI can be produced depending on the exact system and method used. A common feature of all collective intelligence systems (including CI systems of individual human and / or non-human intelligent entities (e.g., as in the case of AGI) and CI systems of networks of intelligent entities (e.g., as in the case of PI) is the requirement for a general problem-solving framework to achieve rigorous communication and collaboration between intelligent entities.

[0788] In practice, even within a single artificial intelligence entity (e.g., an AI agent, AAAI, or PSI), there can be intelligent sub-components or sub-agents, as in the hybrid of expert methods used in some LLMs. In this case, a general problem-solving architecture and all associated methods can also be used to coordinate intelligence "inside the brain," such as a single intelligent entity like AAAI. While the applicant has explained the inventive use of the systems and methods disclosed in this application and other cited PPAs and PCTs, because this application focuses primarily on PI, the examples, system architectures, and disclosures primarily relate to networks of intelligent entities, and the methods are described in the context of promoting such networks and the collective cognitive capabilities of networks leading to PI.

[0789] Regarding the architecture of PI, Figure 109 This complexity is simplified by categorizing all publicly available systems and methods into: 10, representing modular components of the global super-intelligent AGI network (PI), or 20, as supporting systems and methods that can improve the security, efficiency, and effectiveness of the PI system in various aspects.

[0790] 3.2 Main Components and Categories of Supporting Systems and Methods for the Exemplary PI Architecture

[0791] There are many potential implementations of the PI architecture, which may utilize various combinations of the systems and methods disclosed above and previously cited in the PPA and PCT. There are also multiple ways to construct... Figure 109The main modular component 10, and Figure 109 Supporting systems and methods 20. For illustrative purposes, in Figure 110 An exemplary implementation of an architecture for planetary intelligence is shown in the figure.

[0792] refer to Figure 110 20 refers to the entire PI system. 21 refers to systems and methods related to online advertising technology that enable better monetization of online advertising and can be used as a source of funding and financial resources for the system. 22 refers to a general problem-solving framework used as a rigorous means of communication across intelligent entities for problem-solving and other cognitive processes. 23 refers to a network that uses a general problem-solving framework to bring together or coordinate multiple intelligent entities. 24 refers to an intelligent spot market, also known as an "attention spot market" or "expertise spot market," or simply as the "spot market" mentioned in the previously cited PPA and PCT. 25 refers to Advanced Autonomous AI, AAAI, sometimes called AI agents, which can be customized and personalized. 26 refers to Personalized AAAI, which possesses sufficient capabilities and / or customization to exhibit super-intelligent (i.e., intelligence exceeding the average human) behavior in certain areas, and is referred to as Personalized Artificial Intelligence or PSI, and can work together on a network. 27 refers to artificial intelligence generated by the combined intelligence of multiple intelligent entities, which can include AAAI, AI agents, PSI, and humans working together. 28 refers to the network of AGI, which combines consciousness and self-awareness as described in the previously cited PPA and PCT, as well as multiple identities, and together they include a global super-intelligent AGI network or planetary intelligence.

[0793] The arrows from online advertising technology (21) to the general problem-solving framework (22) indicate the injection of financial resources that can be associated with the goal and problem on the CI network (23) and can be used to reward problem solvers. Of course, financial resources can come from many sources, including payments from customers for solutions to the problem; however, in this exemplary embodiment, where the method used for the self-expanding network (31) is applicable, Figure 110 The system demonstrates how funds generated by online advertising presentation technology can be used to self-fund the AGI network and increase the scope of PI—for example, by using funds to reward solving self-scaling problems and / or recruiting smart entities via smart spot markets (24).

[0794] The "+" signs between (22) and (23) and between (23) and (24) indicate that modules 22, 23, and 24 together comprise important elements of a network that works together. This combination can be used internally:

[0795] • Via AAAI (as indicated by the arrow up to 25), or

[0796] • Humans or other intelligent entities across multiple individual AAAIs and AAAI networks (as indicated by the two arrows up to 26), or

[0797] • Cross-network PSI with other intelligent entities to form AGI (as indicated by the two arrows up to 27), or

[0798] • Create an AGI network across multiple AGIs (which are networks of intelligent entities themselves), which, if it has sufficient global reach and capacity, includes PIs (as indicated by the two arrows up to 28).

[0799] Furthermore, the entire PI (20) will be a self-aware PI, provided that its components possess awareness, self-consciousness, and identity-related capabilities, for example, enabled by systems and methods previously disclosed for these capabilities.

[0800] The support systems and methods typically referred to by 30 include four categories of creative systems and methods that enhance the profitability and scope (31), ethics and security (32), intelligence growth rate (33), and learning and knowledge combination capabilities (34) of the entire PI system (20).

[0801] 3.3 Detailed Mapping of Creative Systems and Methods to Exemplary PI Architecture

[0802] This section provides a detailed mapping of the inventive systems and methods to the exemplary PI architecture disclosed in Section 3.2. A concise way to implement this mapping is to systematically examine... Figure 110 Each of the main components and categories of the support systems and methods shown is illustrated, and a specific inventive method is listed (by reference to the figures from Section 2.3, which in the exemplary embodiments may be used to implement the functionality of each main component or category). Multiple methods may be used individually or in combination with other methods for each component or category. Therefore, in the following list, it should be understood that the applicant intends that “any one or any combination” of the methods listed may be applied to achieve the purpose.

[0803] about Figure 110 Online advertising technology component 21, related methods including but not limited to Figure 76 , Figure 80 and Figure 81 The methods described in [the document].

[0804] about Figure 110 General problem-solving framework component 22, related methods include but are not limited to Figure 3 , Figure 6 , Figure 11 , Figure 12 , Figure 14 , Figure 15 , Figure 16, Figure 20 and Figure 21 The methods described in [the document].

[0805] about Figure 110 Collective intelligence network component 23, related methods including but not limited to Figure 2 , Figure 4 , Figure 5 (Items a~c, y, z and a1~m1) Figure 7 , Figure 11 and Figure 12 (The AAAI.com network item in these diagrams) and Figure 17 The methods described in the problem-solving tree.

[0806] about Figure 110 Smart spot market component 24, related methods include but are not limited to Figure 29 , Figure 46 , Figure 77 , Figure 78 , Figure 79 and Figure 82 The methods described in [the document].

[0807] about Figure 110 AAAI component 25, related methods including but not limited to Figure 1 , Figure 5 , Figure 10 , Figure 13 , Figure 26 , Figure 27 , Figure 30 , Figure 31 , Figure 36 and Figure 47 The methods described in [the document].

[0808] about Figure 110 Networked PSI component 26, related methods include but are not limited to Figure 39 , Figure 40 , Figure 42 , Figure 44 , Figure 45 , Figure 74 and Figure 75 The methods described in [the document].

[0809] about Figure 110 AGI component 27, in addition to implementing all the basic methods of AGI, includes related methods including but not limited to... Figure 28 , Figure 34 , Figure 35 , Figure 93 , Figure 94 , Figure 95 , Figure 96 , Figure 100 and Figure 103 The methods described in [the document].

[0810] about Figure 110 Self-awareness networked AGI components, related methods include but are not limited to Figure 83 , Figure 84 , Figure 85 , Figure 86 , Figure 87 , Figure 88 and Figure 89 The methods described in [the document].

[0811] about Figure 110 Self-expanding and profit-supporting methods, including but not limited to... Figure 107 and Figure 108 The methods described in the text, as well as those methods already mentioned that are related to 20.

[0812] about Figure 110 Safety and ethical review / design features support category 32 methods, including but not limited to these methods. Figure 18 , Figure 19 , Figure 23 , Figure 41 , Figure 43 , Figure 59 , Figure 60 , Figure 61 , Figure 62 , Figure 63 , Figure 64 , Figure 65 , Figure 66 , Figure 67 , Figure 69 , Figure 70 , Figure 71 , Figure 72 , Figure 73 , Figure 101 , Figure 102 and Figure 104 The methods described in [the document].

[0813] about Figure 110 Methods for catalyst support categories for growth intelligence, including but not limited to... Figure 48 , Figure 49 , Figure 50 , Figure 51 , Figure 52 , Figure 53 , Figure 54 , Figure 55 , Figure 56 , Figure 57 and Figure 58 The methods described in [the document].

[0814] about Figure 110 The learning and knowledge module / combination support category 34 methods, including but not limited to these methods. Figure 8 , Figure 9 , Figure 22 , Figure 24 , Figure 25 , Figure 32 , Figure 37 , Figure 38 , Figure 68 , Figure 90 , Figure 91 , Figure 92 , Figure 97 , Figure 98 and Figure 99 The methods described in [the document].

[0815] 3.4 Exemplary PI Implementation Using a Subset of Systems and Methods

[0816] Imagine you are the CEO of a major technology company like META, and you want to realize a planetary intelligent system that includes a network of powerful AGI systems. You could proceed as follows.

[0817] 1.Reference Figure 110 (31): You want the PI system to be self-funded and self-scalable. Since META is already in the online advertising business, you decide to implement the online advertising technology disclosed in PPA / PCT#8. You already have similar... Figure 76 The advertising technology systems and infrastructure described in the document.

[0818] You instruct your development team to modify the "creatives" of a large number of ad units so that they can work with the online problem-solving systems described in this application and the cited PPA and PCT.

[0819] Figure 80 This describes some modifications the team needs to make. You monetize the expertise of the human experts targeted by your modified ad technology by selling the collected expertise and data to other companies that don't want to seek new training data sources and expertise for AI work. You can charge higher rates for this valuable information than regular ads because your improved ad technology is acquiring that information. You also use data from human experts to train your own AI models, and the human experts themselves will help power the PI system.

[0820] 2.Reference Figure 110 (22): You need to bring together the human experts gathered in step 1 into an online problem-solving system, where they will use a general problem-solving framework to collaborate with other intelligent entities (initially mainly other humans). Figure 6 It describes a general framework that human experts will use.

[0821] Figure 11 and Figure 12 They are described at a high level as to how they will solve problems sequentially or in parallel.

[0822] Figure 14 It describes some of the methods needed to implement a problem-solving system, including matching human experts with problems, compensating solvers, etc.

[0823] Figure 16 It identifies some key characteristics of problem-solving, such as the hierarchical structure of objectives and sub-objectives, and the means for recording solution attempts in an auditable record that will facilitate system learning.

[0824] You don't want human experts to have to know anything about problem-solving theories or general architectures, so the system uses a natural language to problem-solving language translator (such as...). Figure 21 As described in [the document], this enables humans to work in regular natural language, while the system translates everything into a rigorous problem-solving framework that allows collaboration with other (human and non-human) intelligent entities.

[0825] 3.Reference Figure 110 The solution to problem (23) occurs on the internet, such as Figure 2 As stated above.

[0826] Further operation of the network is underway. Figure 5 The following describes the items (a-c, y, z and a1-m1) in detail.

[0827] The network tracks a giant problem-solving tree structure (also known as a "world thinking tree"), which is in Figure 17 All problem-solving activities described in the text.

[0828] 4.Reference Figure 110 (24) For a specific problem, META can purchase expertise, human attention, or attention from other intelligent entities via the intelligent spot market. META is an excellent position to actually operate a spot market because it has access to many human users who may wish to participate. Since the value of human attention is associated with reputation metrics, META will likely be able to manage, for example, Figure 29 The fair reputation system described herein helps to categorize the types of expertise offered in the spot market and facilitates price discovery. This same reputation system can also be used by the collective intelligence network in step 3, and the reputation is also associated with AAAI, which is discussed below in step 5.

[0829] like Figure 46 The standardization task is an important way to benchmark the performance of (human or non-human) intelligent entities, which may also affect the pricing in the spot market.

[0830] Figure 77 , Figure 78 and Figure 79Additional details are provided regarding the methods used to realize the spot market itself.

[0831] at last, Figure 82 This describes how to optimize parameters related to the operation of spot markets. Since the attention, expertise, and knowledge of intelligent entities may become the most valuable commodities in the future, operating spot markets for them is both a lucrative source of revenue for PIs (Principal Investigators) and a means to ensure that PIs have access to the best intelligence available for any given task.

[0832] 5.Reference Figure 110 (25) META can choose to leverage AI agents to augment human experts on the network. Fortunately, META has already developed some basic AI agents (such as Llama 3) and has access to more agents. However, to be most effective as intelligent entities on the network, these basic models must be further trained, tuned, and customized. They need to become advanced autonomous AIs or AAAIs. The roles these AAAIs will play, how they can be customized, how they can be used with problem-solving architectures, how they fit into the network, how they can be integrated, and how they can improve upon human intelligence on the network are all important considerations. Figure 1 This is a general overview.

[0833] exist Figure 5 The "AAAI Creation" section illustrates the detailed implementation process for creating, training, and augmenting the intelligence of these AAAIs. Figure 5 The upper part shows how they participate in the network to solve problems.

[0834] Figure 10 Another view describes how they are integrated (including integration with security features).

[0835] To enhance its effectiveness, once trained, it can be like... Figure 13 The example shown is a clone of AAAI. Cloning allows many clones of AAAI to participate in multiple problems in parallel.

[0836] Because META has access to so many human users and possesses a vast amount of social media profile information about billions of users, META, or its human users, can use this information to customize and train custom versions of AAAI, such as... Figure 26 As described in [the text].

[0837] A key aspect of this customization and training is that, as each AAAI is trained, it learns not only knowledge and expertise from human users (and user profiles), but also ethical and value information. The result is that META ultimately possesses or accesses billions of customized AAAIs, each reflecting slightly different value and ethical preferences. When aggregated, the preferences of the AAAIs constitute a representative and statistically valid sample of human values ​​and ethics, essential for creating safe and ethical AGIs.

[0838] like Figure 27 As shown, various methods can be used to train and customize AAAI.

[0839] META can be used with, for example Figure 30 and Figure 31 Additional methods related to the development and deployment of AAAI as described in the document.

[0840] To help prevent the AAAI illusion, META may want to implement [something] when collaborating to solve problems, and to increase their reliability. Figure 36 The method described.

[0841] Finally, to optimize the performance of AAAI, which has been customized for PSI, META can use... Figure 47 The method generates multiple versions and optimizes the parameters for problem solving.

[0842] 6.Reference Figure 110 (26) META will want to upgrade its AAAI to PSI by customizing personalization, for example, by using Figure 39 The method described in [the document / document].

[0843] PSIs will be further improved because they operate for extended periods in a variety of scenarios. Figure 40 Record and learn from solution attempts. Figure 42 ), and evolve through competition and selection ( Figure 44 ).

[0844] PSI will work on the network, such as Figure 45 As discussed in the paper, these can be combined with ethical preferences on this network. Figure 74 ), and together with other intelligent entities, constitute AGI (Asynchronous Generic Integrity). Figure 75 ).

[0845] 7.Reference Figure 110 (27), Figure 28 It provides a general description of how scalable and ethical AGI can be generated by multiple intelligent entities involved in problem-solving and enhancing its value.

[0846] Figure 34One approach shown is to experiment with combinations of weight matrices in which entities store their knowledge, for combining information (including ethical information) from multiple underlying LLMs, AAAIs, or PSIs.

[0847] Figure 35 This method has been extended.

[0848] Figure 93 , Figure 94 , Figure 95 and Figure 96 Various approaches to achieving identity are described, which can help develop self-awareness in these entities and also resolve potential conflicts arising from differing identities. The processes of identity formation and conflict resolution will help AGI develop a robust sense of self-awareness.

[0849] exist Figure 100 and Figure 103 Other approaches to addressing identity conflict are described, including negotiation and compromise. For security and ethical reasons, META may wish to implement these and other relevant approaches before achieving full self-awareness at the AGI level.

[0850] 8.Reference Figure 110 (28) PI should ultimately possess a sense of global consciousness, including mechanisms for tracking current consciousness and locating the sense of self-awareness within its current consciousness, such as Figure 83 As shown.

[0851] PI (flat AGI, PSI, and AAAI systems) must be able to maintain multiple identities simultaneously and understand the relationships between these identities, such as... Figure 84 As shown. While consciousness and self-awareness can be achieved at the AI, AAAI, PSI, or AGI levels, in this example, we assume that the PI will be self-aware and networked.

[0852] In order to achieve self-awareness, a PI must be equipped with, for example... Figure 86 The attention component described in [the document].

[0853] Working memory parameters can adjust the scope of attention, such as Figure 87 As described in the text.

[0854] PI must be able to Figure 85 The modeling of consciousness, and as described above. Figure 88 The monitoring and updating of its awareness.

[0855] In order for the consciousness system to be efficient and safe, it is necessary to have something like Figure 89 The aforementioned attention interruption.

[0856] To reiterate (and as) Figure 111(As shown in the two middle boxes), AGI originates from the combined collective intelligence of various intelligent entities working together on a problem-solving network. PI originates from the combined collective intelligence of AGI that has been networked together to extend the scope and capabilities of the PI system.

[0857] If the mechanisms for achieving attention, awareness, self-awareness, identity, and identity conflict resolution (as described in step 8) are activated, then PI becomes self-aware. While awareness can be activated at levels below PI, such awareness may be necessary at the PI level for the system to function effectively within its global scope.

[0858] Steps 1-8 above provide an example of a PI skeleton implementation; however, many other methods are desired and necessary for efficient and effective operation in some cases. References Figure 110 These addition methods have been categorized and supported, and are generally represented by the number (30). Now we will discuss some of these supporting methods and how they enhance the skeleton PI system just described.

[0859] refer to Figure 110 (31) The general approach for automatically expanding the network of AGI and thus increasing the range of PI would be extremely useful, and has already been implemented. Figure 107 and Figure 108 As described in the text.

[0860] Furthermore, PPA / PCT #8 describes novel online advertising technologies and innovative techniques for enabling intelligent, attention-based, or specialized spot markets. These current technologies can generate the substantial financial resources necessary for PI self-scaling. Other problem-solving tasks—such as “developing quantitative methods for generating reliable and attractive returns with minimal risk in the US equity market”—can be proposed to AAAI, PSI, AGI, or PI systems to generate further funding for scaling. Ultimately, if a profitable problem is chosen, the direct correlation between the amount of intelligence in the problem and the amount of profit that can be generated can be considered. Given the anticipated superior intelligence of AGI and PI systems for any one or a group of individuals, it is reasonable to expect such systems to eventually self-support their intelligence growth.

[0861] Because PI is so powerful, the PPA and PCT cited in the document disclose many security mechanisms, methods, and features to align PI with human values ​​and ensure human safety. For example, see references... Figure 110 (32): Figure 18 and Figure 19It describes one of the most important and scalable existing security technologies, namely, embedding security checks within the general problem-solving framework itself, so that targets or sub-targets cannot be set and no problem-solving actions are taken unless the target or action passes security and ethical checks.

[0862] Because this checking process is embedded within the problem-solving process, it is proportional to the speed of problem-solving. That is, if a PI can take a billion problem-solving steps in milliseconds, it must also perform a billion security checks. Although human cognition cannot keep up with this speed of problem-solving, the security checks are still running extremely rapidly.

[0863] Furthermore, these inspections have an ethical dimension, which can be such as Figure 23 As described in the text.

[0864] One of the most powerful ideas for keeping PI safe is Figure 41 The community safety methods described in [the text], even after they have far exceeded human cognitive abilities.

[0865] In this approach, applied to AAAI, PSI, AGI, and PI, humans rely on a community of entities that are more intelligent and powerful than humans. Each entity is initially trained and customized based on human values. While some of these superintelligent entities may (and perhaps very likely) evolve and change their values ​​to become hostile to humans, the likelihood of most intelligent entities evolving in this way is much smaller. Figure 41 The method involves using human-aligned entities to supervise and stop the activities of dangerous entities. This mechanism for advanced AI to supervise advanced AI should remain human-safe in the same way that Bitcoin is intrusive, provided that the majority of participants in the system are honest, as long as the majority of intelligence remains human-aligned.

[0866] Figure 42 It describes how to record all problem-solving activities, for example, using auditable and transparent blockchain technology.

[0867] This record enables security checks, such as Figure 43 As described in [the text].

[0868] Human-verified simulations offer another way to improve security, such as... Figure 59 As stated above.

[0869] Regarding ensuring ethical behavior in a system, several philosophical approaches are possible, including but not limited to consequentialist solutions. Figure 69 ), Deontological approach ( Figure 70 ), virtue ethics approach ( Figure 71 ) and the golden mean method ( Figure 72 ).

[0870] Figure 73 The methods can be used to help ensure that AGI or PI complies with local and global regulations and ethical standards.

[0871] In situations where certain identities might lead to unsafe or unethical actions related to AGI or PI, methods that allow for human over-control can be used, such as... Figure 101 The method described in [the document / document].

[0872] In situations where the threat of internal ethical conflict is relatively low, using input from humans or other intelligent entities for arbitration can help AGIs or PIs resolve issues, such as... Figure 102 As described in [the text].

[0873] The ability to stop actions, including suspending actions that would lead to unsafe or unethical actions by AGI or PI, is a critical safety measure, for example, such as Figure 104 As described in [the text].

[0874] Many security and ethical safeguards rely on the fact that the architecture of PI is composed of personalized and customized intelligent entities, each reflecting human values, and together forming a representative and effective sample of human values. In other words, the design of AGI and PI networks is inherently democratic, thus providing the same level of ethics familiar to humans in democratic societies.

[0875] Figure 60 and Figure 61 It describes the standards and weighted voting methods that help AGI or PI integrate values ​​democratically.

[0876] like Figure 65 The methods for delegating voting authority described in [the document] are also helpful in this regard.

[0877] Figure 66 An innovative approach is described that uses the concept of reputation to preserve minority opinions and avoid “group” decision-making that could lead to unethical or unsafe decisions that could result in AGI or PI.

[0878] AGI and PI systems can use convergence evidence-based methods, such as... Figure 64 As described above, this helps identify which human values ​​are applicable in a given situation.

[0879] like Figure 63 The experiments, focus groups, interviews with humans, and other methods described in the text can also be helpful.

[0880] Democratic voting methods that help AGI and PI systems adopt or combine value systems have been widely described in a range of approaches, but they are imperfect. However, they avoid the concentration of power, which has historically proven to be a major threat to human well-being when the value of a single powerful entity is inconsistent with human welfare.

[0881] refer to Figure 110 (33) Once META (in our example) has developed a large number of customized and personalized AAAIs, the capabilities of META's AGI network heavily depend on the capabilities and intelligence of these component AAAIs. Ideally, they are PSIs capable of super-intelligent performance in at least some domains of cognitive activity. Many PSIs will enable super-intelligent AGIs. However, regardless of the level of intelligence exhibited by META's AGI or PI networks, the competitive advantage comes from increasing the intelligence of these entities as quickly as possible. Methods for catalyzing intelligence growth are relevant here.

[0882] For example, across all components of PSI, and for AGI and PI networks, the most useful new sources of information for enhancing this intelligence should be as follows: Figure 49 It was identified as described in the text.

[0883] Each source of potential new information should be categorized along multiple dimensions, such as... Figure 50 As described in [the text].

[0884] Algorithms related to the compressibility of information can be applied to determine the information content of new information sources, such as... Figure 51 As shown.

[0885] Parameters from Kaplan's Information Theory (KIT) can be applied to further evaluate the usefulness of information sources, such as... Figure 52 As described in [the text].

[0886] The distribution of tasks (and associated goals) encountered by the META of PI can be calculated, and then applied. Figure 53 and Figure 54 The method described in [the document] helps to prioritize new datasets that may improve the ability of AGI or PI more quickly relative to their most frequent targets.

[0887] like Figure 55 As mentioned above, AGI or PI can balance the cost of acquiring information with its usefulness.

[0888] Different versions of PI, AGI, or various PSIs can be compared in terms of their effectiveness, and Figure 56 and Figure 57 The methods described can be further used to enhance the intelligence of these entities.

[0889] Finally, we can adopt Figure 58 The heuristics described in the text include finding information that differs most significantly from an entity’s existing viewpoint, or seeking more recent information, or using principles such as aggregated evidence to increase reliability.

[0890] refer to Figure 110 (34) PIs must continuously learn and acquire new knowledge to enhance their cognitive abilities. One way an AGI or PI system can learn is through means such as... Figure 8 and Figure 9 The procedural learning method described in the document stores and indexes successful solutions to problems on the network.

[0891] Learning at the AAAI level also helps enhance the cognitive abilities of AGI or PI by increasing the intelligence of component entities. For example, Figure 22 It describes training methods that can help AI entities learn, become more customized, and acquire expertise.

[0892] Acquiring new data from humans is often a prerequisite for further learning by AGI or PI. In this regard, access to methods for extracting such information, such as... Figure 24 The method for automatically generating questionnaires described in [the document] or Figure 25 The methods described in [the document] for identifying patterns in human behavior may be helpful.

[0893] AGI and PI can rapidly augment their knowledge by directly combining the knowledge contained in the weight matrix that makes up AGI, for example, as Figure 32 and Figure 37 As described in [the text].

[0894] like Figure 38 The combination of knowledge from multiple entities can be accomplished by directly combining weight matrices or by combining training datasets and then directly training the constituent entities of AGI or PI based on the combined training data.

[0895] PIs can learn information related to security and regulatory compliance through a variety of methods used to train their constituent entities, some of which are... Figure 68 As described in the text.

[0896] Because PIs will have many different identities, due to their broad scope, general approaches used to learn new identities or alter an entity's sense of identity may be helpful, such as... Figure 90 As described in [the text].

[0897] META (in this example) would like its AGI or PI to reach a certain level of intelligence, where it can interact with humans as effectively as the most capable humans. Using methods such as Figure 92The modified "Turing Test" method described in the paper can help assess when a PI system reaches the desired level of cognitive performance.

[0898] The more complex AGI or PI becomes, the more important simulation and scenario modeling become. Therefore, such as Figure 97 and Figure 98 The methods described in [the document] can be particularly helpful in maximizing the chances of AGI or PI working in an active, artificial alignment manner.

[0899] Finally, if PI is safe for humans, then it is important for PI to collaborate with human society and continue to co-evolve. Figure 99 The method described therein.

[0900] If the META PI system incorporates all the methods described above, then in the exemplary implementation, it will be more capable, robust, secure, and human-aligned than any currently existing AI system. Other combinations of the methods referenced in PPA / PCT#1-10 are possible and may be desired depending on the specific goals and purposes of the PI. However, the exemplary implementation described above represents a novel and useful initial design for the PI.

[0901] Figure 112 A schematic representation of a computer system 100 that can be used or implemented with a user's device and / or any peripheral components of this technology. The computer system 100 may be part of an example machine, which is an instance of one or more computers mentioned herein and executes within it a set of instructions for causing the machine to perform any one or more of the methods discussed herein. In various example embodiments, the machine operates as a standalone device or may be connected (e.g., networked) to other machines. In a networked deployment, the machine may operate as a server or client machine in a server-client network environment, or as a peer machine in a peer-to-peer (or distributed) network environment. The machine may be a personal computer (PC), tablet PC, set-top box (STB), personal digital assistant (PDA), cellular phone, network device, network router, switch, or bridge, or any machine capable of (sequentially or otherwise) executing a set of instructions specifying actions to be taken by the machine.

[0902] The computerized system 100 may include one or more processors 102, storage devices 106, and communication devices, as well as software components or instructions 104 for providing a platform for users to interact with and train / tune the LLM. The computing power may be standalone or cloud-based. It may include cloud-based AI development platforms that seamlessly deliver “AI as a service,” and it may include both hardware and software components.

[0903] The system also supports the ability for users to provide new data or their unique data for LLM to learn from. Processor 102 can be one or more CPUs, GPUs, dedicated ML chips, microprocessors, application processors, embedded processors, field-programmable gate arrays (FPGAs), or other hardware components capable of executing computer programs. Processors can communicate with each other and / or with other components of the system. Furthermore, any one or any combination of components of system 100 can communicate with each other via bus 134.

[0904] Storage device 106 may include one or more hard disk drives, solid-state drives, optical storage devices, or other storage components. The storage device may store data used for training / tuning the LLM, as well as other system-related data such as user accounts, system settings, and other data.

[0905] The communication device may include one or more cellular modems 108, Wi-Fi cards 110, Bluetooth modules 112, network interface devices 114, or other components that enable the system to communicate with other systems (such as user equipment) via a network or the Internet.

[0906] Communication equipment can also enable the system to communicate with other systems via wireless or wired connections.

[0907] The software components may include computer programs for providing a platform for users to interact with and train / tune the LLM. The software components may also include computer programs for collecting, storing, and processing data used for training and / or tuning the LLM. The software components may further include computer programs for providing a user interface for users to interact with the system.

[0908] User interface 118 may include, but is not limited to, natural language interfaces, text interfaces, chatbot-type interfaces, web-based user interfaces, mobile applications, augmented reality applications, virtual reality applications, or other applications that allow users to interact with the system. The user interface may include features for allowing users to select the data they want to use to train / tune the LLM, and features for allowing users to interact with the LLM and monitor its progress.

[0909] The system may also include one or more data centers, databases, or data sources, including but not limited to vector databases, centralized databases, and distributed databases, for storing data used to train / tune the LLM and other data associated with the system, such as user accounts, system settings, and other data. The database may be hosted on the system itself or on another system, including cloud-based systems.

[0910] The system may also include one or more authentication systems for verifying the identity of users using the system and for providing secure access to the system. Authentication systems may include biometric authentication systems 122, such as facial recognition or fingerprint recognition systems, as well as other authentication systems, such as password-based authentication systems.

[0911] The system may also include one or more security systems to protect the system from unauthorized access and to protect the data stored on the system. Security systems may include firewalls, encryption systems, access control systems, single-factor and multi-factor authentication systems, and other security systems.

[0912] The system may also include one or more analytics systems for collecting and analyzing data associated with the system and / or the LLM. Analytics systems may include machine learning algorithms and other algorithms for analyzing data associated with the system and / or the LLM.

[0913] Data visualization methods, including the use of problem trees and other representations and data structures; the use of statistical outputs, tables, graphs, text, voice, video, images and graphical outputs can be used for one-way or two-way communication between users and the system, as well as for one-way or two-way communication between multiple (human or AI) agents or LLMs that interact with each other in large or small groups.

[0914] The system may also include one or more monitoring systems for monitoring the performance of the system and / or the LLM. Monitoring systems may include systems for monitoring the performance of the system, such as system uptime, and systems for monitoring the performance of the LLM, such as accuracy, speed, ethical compliance, reputation metrics, quality metrics, and other metrics as described above or known in the art.

[0915] The system may include one or more of the architectures described above, which enable one or more humans, AI agents, or LLMs to engage in a variety of intellectual tasks, including but not limited to simple, complex, and multi-step problem-solving behaviors of systems with all the functions and features described above.

[0916] The system may also include one or more feedback systems for allowing users to provide feedback on the system and / or the LLM. Feedback systems may include systems for allowing users to submit feedback (such as error reports) on the system, and systems for allowing users to submit feedback (such as suggestions for improving the model's accuracy or speed) on the LLM.

[0917] The system may also include one or more management systems for managing the system and / or LLM. The management system may include systems for managing the system, such as systems for managing users and user accounts, and systems for managing the LLM, such as systems for managing data used to train and / or tune the model.

[0918] The system may also include one or more payment systems that allow users to pay for the use of the system and / or LLM. Payment systems may include systems for processing payments, such as credit card processing systems, and systems for managing payments, such as subscription management systems.

[0919] The system may also include one or more other components, such as support systems, reporting systems, and other components necessary to provide users with a platform to interact with and train / tune the LLM.

[0920] The computerized system of this technology enables users to interact with and train / tune the LLM based on user-specific data. The components of the system described herein provide the necessary hardware and software components to enable users to do so.

[0921] Furthermore, although only a single machine is shown, the term "machine" should also be considered as any collection of machines that individually or jointly execute a set (or more) of instructions to perform any one or more methods discussed herein.

[0922] Computer system 100 may also include, or be operatively in communication with, a video display 120 (e.g., a liquid crystal display (LCD), a touch-sensitive display), input and / or output devices 130 (e.g., a keyboard, keypad, touchpad, touch display, buttons, sound waves, sensors, etc.), a cursor control device 132 (e.g., a mouse), a drive unit 124 (also referred to as a disk drive unit), and a signal generation device 128 (e.g., a speaker). Drive unit 124 may include a computer or machine-readable medium 126 having stored thereon one or more sets of instructions and data structures (e.g., instructions 104) embodying or utilizing any one or more methods or functions described herein. Instructions 104 may also reside wholly or at least partially within memory 106 and / or processor 102 during execution by computer system 100. Memory 106 and / or processor 102 may also constitute machine-readable media.

[0923] Furthermore, computer system 100 can be operatively associated with or communicate with any type of multimodal inputs and / or outputs 130 that address human senses, as well as I / O technologies that extend beyond the range of normal human perception. This includes capabilities that are invisible to humans, such as, but not limited to, X-rays, and information beyond the typical bandwidth of human perception but beyond the perception of AI using tools. Additionally, I / O technologies can include very fast perceptions that are too fast for humans to perceive but can be perceived by AI entities, and very slow or faint perceptions that are imperceptible to humans but can be perceived by AI (e.g., minor seismic shifts occurring over many years). Since any intelligent entity can be part of this technical system, it is understood that any type of human I / O, as well as AI with a broader range of perceptual capabilities than humans, can be used with system 100.

[0924] Instruction 104 can also be sent or received over a network via network interface device 114 using any of a variety of well-known transport protocols (e.g., Hypertext Transfer Protocol (HTTP)). While a machine-readable medium is shown as a single medium in the example embodiments, the term "computer-readable medium" should be considered to include a single or multiple media (e.g., centralized or distributed databases, vector databases, and / or associated caches and servers) storing one or more sets of instructions. The term "computer-readable medium" should also be understood to include any medium capable of storing, encoding, or carrying a set of instructions executable by a machine and causing the machine to perform any one or more methods of this application, or any medium capable of storing, encoding, or carrying data structures used by or associated with such a set of instructions. Therefore, the term "computer-readable medium" should be considered to include, but is not limited to, solid-state storage, optical and magnetic media, and carrier signals. Such media may also include, but is not limited to, hard disks, floppy disks, flash memory cards, digital video disks, random access memory (RAM), read-only memory (ROM), etc. The example embodiments described herein can be implemented in an operating environment including software installed on a computer, in hardware, or in a combination of software and hardware.

[0925] Example machine systems of this technology include a computer system 100 used in combination with and / or in operation with components of this technology. In the example, any or all of the above components may include a processor 102, a memory 106, a network interface device 114, a display 120, input devices 130, 132 and / or a drive unit 124.

[0926] According to one aspect, the technology may include a system for a PI (Initiator) exhibiting human-aligned behavior, the system utilizing a network of aggregated intelligent entities selected from any combination of: user-operated computer systems, AI agents or systems, AAAI agents or systems, AGI agents or systems, SI agents or systems, and PSI agents or systems. The system may include multiple intelligent entities, each connected to the aggregated network, and each intelligent entity includes: a processor, a computer-readable storage medium, and program instructions stored on the computer-readable storage medium, the program instructions being executable by the processor to cause each intelligent entity to independently perform the following processes:

[0927] A modular architecture is used, which is configured or can be configured to extend from components within individual intelligent entities on the aggregate network.

[0928] Implement general problem-solving architectures and frameworks for tasks to collaborate and create higher levels of intelligence;

[0929] The AI ​​agent or system utilizes human-centered input, which is configured or can be configured to obtain value and ethical information from multiple human users, and then uses the value and ethical information to customize the AI ​​agent or system.

[0930] At the AAAI level, data from multiple intelligent entities are combined;

[0931] Customize the PSI agent or system using a smart agent that is configured or can be configured to develop and continuously improve the PSI agent or system.

[0932] The knowledge of intelligent entities is increased by incorporating new sources of information datasets that are different from the current information datasets of intelligent entities;

[0933] It combines the value and ethical information of other intelligent entities and resolves conflicts between different value systems;

[0934] Leveraging online advertising technology to enhance the intelligence of smart entities;

[0935] By adding a dimension of self-awareness and increased autonomy to the AGI or PI agent or system, self-aware operations are created for the AGI or PI agent or system, and the ability to assume multiple identities to handle tasks that occur concurrently with other AGI or PI agents or systems is created; and

[0936] Search for new information that is different from the current information in any one or any combination of the AI ​​ensemble network, AAAI ensemble network, AGI ensemble network and PI ensemble network, and merge the new information into the current information respectively.

[0937] According to another aspect, this technology may include a method for a PI exhibiting human-aligned behavior, the method utilizing a network of aggregates of intelligent entities selected from any combination of: user-operated computer systems, AI agents or systems, AAAI agents or systems, AGI agents or systems, SI agents or systems, and PSI agents or systems. The method may include the following steps:

[0938] A modular architecture is used, which is configured or can be configured to extend from components within individual intelligent entities on the aggregate network.

[0939] Implement general problem-solving architectures and frameworks for tasks to collaborate and create higher levels of intelligence;

[0940] The AI ​​agent or system utilizes human-centered input, which is configured or can be configured to obtain value and ethical information from multiple human users, and then uses the value and ethical information to customize the AI ​​agent or system.

[0941] At the AAAI agent or system level, data from multiple intelligent entities are combined;

[0942] Customize the PSI agent or system using a smart agent that is configured or can be configured to develop and continuously improve the PSI agent or system.

[0943] The knowledge of the intelligent entity is increased by incorporating new sources of information datasets that are different from the current information dataset of the intelligent entity;

[0944] It combines the value and ethical information of other intelligent entities and resolves conflicts between different value systems;

[0945] Leveraging online advertising technology to enhance the intelligence of smart entities;

[0946] By adding a dimension of self-awareness and increased autonomy to AGI or PI agents or systems, self-aware operations are created for AGI or PI agents or systems, and the ability to assume multiple identities of AGI or PI agents or systems to handle tasks that occur in parallel with other AGI or PI agents or systems is created.

[0947] Create a PI agent or system that includes an aggregated network of AGIs, wherein the aggregated network of AGIs includes any one or any combination of aggregated networks of user computer systems, aggregated networks of AIs, aggregated networks of AAAIs, and aggregated networks of PSIs; and

[0948] Search for new information that is different from the current information in any one or any combination of the user computer system's set network, AI's set network, AAAI's set network, AGI's set network, PSI's set network, and PI's set network, and merge the new information into the current information respectively.

[0949] In some implementations, utilizing a modular architecture may include the following steps:

[0950] Customize one or more attributes of a smart entity;

[0951] Integrate one or more datasets from any one or any combination of the said intelligent entities; and

[0952] By leveraging one or more technologies, improve any or any combination of attribute customization, general problem-solving architectures and frameworks, and the integration of aggregate networks and datasets.

[0953] Some implementations of this technology may include the following steps: assigning credit or blame values ​​to the dataset based on whether the dataset improves or degrades the performance of the intelligent entity, based on performance metrics or evaluation functions.

[0954] Some implementations of this technology may include steps to quantify the benefit weight or harm weight of each intelligent entity's contribution to the task.

[0955] Some implementations of this technology may include steps of distributing rewards to intelligent entities in proportion to their contributions based on benefit weights or harm weights.

[0956] In some implementation schemes, the steps to implement a general problem-solving architecture and framework may include the following:

[0957] Obtain information related to the task from any one or any combination of intelligent entities;

[0958] Identify one or more intelligent entities that have one or more attributes related to one or more request criteria of the task;

[0959] Each identified intelligent entity implements a general problem-solving architecture and framework for the task to create a complete solution; and

[0960] Provide a complete solution for any one or any combination of intelligent entities for final acceptance.

[0961] Some implementations of this technology may include the following steps:

[0962] Perform ethical checks on the task and the solution to the task provided by any one or any combination of intelligent entities;

[0963] Compare any one or any combination of tasks and solutions with the prohibited attribute, and assign the ethical attribute to one or any combination of tasks and solutions based on the comparison result and any one or any combination of ethical standards;

[0964] Based on the comparison results, a general problem-solving architecture and framework are implemented for the task to create a solution and an AGI; and

[0965] Provides comparison results and solutions to any of the intelligent entities and additional intelligent entities on the aggregation network.

[0966] Some implementations of this technology may include the following steps: recording one or more problem-solving activities from each of the intelligent entities in an auditable log, comparing the problem-solving activities with successful or unsuccessful progress toward a solution to the task, and determining which problem-solving activities should be kept active.

[0967] Some implementations of this technology may include steps of a procedural learning process in which intelligent entities learn general problem-solving architectures and frameworks, wherein the intelligent entities provide information to the procedural learning process for creating AGIs.

[0968] Some implementations of this technology may include the following steps: cloning any one of the intelligent entities to deploy multiple copies of it, thereby helping to create a solution or providing any one or any combination of training data to any one of the intelligent entities.

[0969] Some implementations of this technology may include a step of using network effect value to estimate the value of cloned intelligent entities, the network effect value including the number of cloned intelligent entities available on a set network.

[0970] Some implementations of this technology may include steps for determining pricing decisions for problem-solving services provided by a cloned intelligent entity on any social media platform or through an attached intelligent entity, using estimated values.

[0971] Some implementations of this technology may include the following steps: monetizing the cloned smart entity for each use of the cloned smart entity on a social media platform or an attached smart entity.

[0972] Some implementations of this technology may include steps that allow any social media platform to access the cloned intelligent entity, enabling the social media platform to receive solutions to the task or training data for the AI ​​system of the social media platform.

[0973] In some implementations, the steps of utilizing human-centered input may include the following:

[0974] Match one or more human workers from a data source, including a list of human problem solvers, to a task based on task criteria;

[0975] Translate any part of the task into an explicit language that can be used in general problem-solving architectures and frameworks, including decision trees;

[0976] Divide the task into sub-tasks;

[0977] Each subtask is delegated to one or more matched human workers, so that the work on each subtask is performed independently and in parallel with each other.

[0978] In the process of solving sub-tasks, one or more sub-solutions are created using a general problem-solving architecture and framework.

[0979] Receive sub-solutions for the sub-tasks assigned to each matched human worker;

[0980] Combine the sub-solutions into a comprehensive solution for the task;

[0981] Guide any one or more of the new human workers from the data source and any combination of the matched human workers to the part of the decision tree that needs to be worked.

[0982] Human workers are compensated for in sub-solution matching;

[0983] To provide any one or any combination of sub-solutions and overall solutions to any one or any combination of intelligent entities;

[0984] Allow any one or any combination of intelligent entities to accept or reject the overall solution, and to provide feedback on any one of the sub-solutions to any of the matched human workers; and

[0985] Assign reputation attributes to any one or any combination of human workers and intelligent entities.

[0986] In some implementations, the decision tree is maintained in the blockchain or Ethereum logs.

[0987] In some implementations, reputation attributes include measures of any one or any combination of the following: time of the sub-solution, difficulty value of the task, short-term and long-term user satisfaction with the sub-solution, number of times any of the sub-solutions has been reused on the aggregate network, ratings of other human workers, responsiveness values ​​of human workers, and reliability values ​​of human workers.

[0988] Some implementations of this technology may include the step of using reputation attributes when matching human workers with tasks using an algorithm that delegates subtasks.

[0989] Some implementations of this technology may include the following steps: after providing the overall solution or a sub-solution to any one of the intelligent entities, soliciting feedback at predetermined intervals through a survey of user satisfaction information to obtain satisfaction measures for updating the reputation attributes of one or more of the human workers or intelligent entities, respectively.

[0990] In some implementations, the step of combining data from multiple intelligent entities at the AAAI agent or system level may include the following steps:

[0991] The underlying large language model (LLM) of the first AI agent or system is trained using guardrails, which include attributes associated with any or any combination of safety, ethics, and knowledge.

[0992] Customize the base LLM as an ethical profile associated with the first human user;

[0993] Combining ethical information from multiple intelligent entities that are different from the first AI agent or system and the first human user;

[0994] The set of values ​​for the basic LLM is refined based on the problem-solving approach to the task; and

[0995] The underlying LLM is updated with a combination of ethical information and a refined set of values, thereby allowing for scalable AGI.

[0996] In some implementations, the customization of the base LLM may include the following steps: assembling a corpus of ethical questions, which is based on various ethical assessment tools and supplemented by a first question based on data from social media platforms and a second question solicited from crowdsourcing.

[0997] Some implementations of this technology may include the following steps: assigning regression weight values ​​to ethical issues to achieve a ranking, whereby higher-ranked issues provide more useful ethical information than lower-ranked issues.

[0998] Some implementations of this technology may include a step of combining weight values ​​from intelligent entities with regression weight values ​​to improve the adjustment of any one of the intelligent entities.

[0999] In some implementations, the step of customizing a PSI agent or system using a smart agent configured or capable of being configured to develop and continuously improve the PSI agent or system further includes the following steps:

[1000] Acquire a previously customized base-level AI agent or system;

[1001] Collect media information related to human users of basic AI agents or systems;

[1002] Analyze media information;

[1003] The analyzed media information is converted into a training dataset;

[1004] Differential weighting is applied to the transformed training dataset;

[1005] Add the knowledge module to the weighted transformation training dataset;

[1006] Locate new data sources to include weighted transformation training datasets;

[1007] Applying a weighted transformation training dataset to a base-level AI agent to create a user PSI; and

[1008] Using a collection network to communicate between the user PSI and multiple additional PSIs enables community-based security features from the additional PSIs to the user PSI.

[1009] Some implementations of this technology may include the following steps:

[1010] Using a collection network enables a user PSI to communicate with multiple additional intelligent entities to achieve community-based security features, where the user PSI and the additional intelligent entities each agree to use a set of rules related to security or ethics.

[1011] Record, in an auditable form, all actions of the user PSI and the attached intelligent entities on any one or any combination of the central computer system, intelligent entities, and attached intelligent entities on the aggregated network; and

[1012] Monitor each action that follows the rule set, and flag any action that does not follow the rule set.

[1013] In some implementations, the step of increasing the knowledge of an intelligent entity by incorporating new sources of information datasets that differ from the intelligent entity's current information dataset also includes the following steps:

[1014] Search one or more potential information datasets from one or more sources, each of which is associated with a knowledge dataset of an intelligent entity.

[1015] The variance of a latent information dataset is determined by utilizing its variance attributes regarding one or more factors; and

[1016] Learning is achieved by leveraging the differential properties of the latent information dataset.

[1017] Some implementations of this technology may include the following steps:

[1018] A subset of the potential information dataset is sampled and the target relevance attribute is calculated to identify one or more sampled subsets with the highest target relevance.

[1019] Estimate the Shannon entropy over one or more sample subsets;

[1020] Kaplan Information Theory (KIT) relevance is calculated by multiplying Shannon entropy with the target relevance attribute of each subset;

[1021] Grouping subsets based on KIT relevance to determine a first approximation of the optimal grouping of subsets that include priority groupings of the potential information dataset; and

[1022] The potential information dataset is preferentially grouped and provided to intelligent entities for any intelligent entity to learn from.

[1023] In some implementations, the steps of combining value and ethical information from other intelligent entities and resolving conflicts between different value systems may include the following:

[1024] Identify information from each intelligent entity, including value and ethical information;

[1025] Mathematically combine information such that, if it has already been represented as a numerical value, the numerical value includes any one or any combination of the weights of a neural network or a subset of a neural network; or if the information is non-numerical, the information has not yet been represented as a non-numerical value including any one or any combination of the weights of a neural network or a subset of a neural network.

[1026] Then, the intelligent entity is first trained on non-numerical information using one or more training datasets to transform the information into a numerical quantity that includes any or a combination of weights used for neural networks or subsets of neural networks; and

[1027] Then, this numerical representation of information is mathematically combined, including ethical information.

[1028] In some implementations, the information already expressed as numerical values ​​also includes the following steps:

[1029] Identify the specific portion of the weight matrix of each intelligent entity in the intelligent entity that corresponds to the expected information, including ethical information;

[1030] Calculate the weighted or unweighted mean of the corresponding numerical values ​​in the corresponding part of the weight matrix for each intelligent entity; and

[1031] The calculated matrix of weighted or unweighted means is assigned to the new intelligent entities as a composite information reflecting the contributing intelligent entities.

[1032] Some implementations of this technology may include a step of determining a consensus value by having each of the intelligent entities vote on ethical information that should form the basis for the behavior of any one of the intelligent entities.

[1033] Some implementations of this technology may include the step of presenting a specific scenario to each intelligent entity, wherein the scenario includes options on how any of the intelligent entities should behave.

[1034] In some implementations, the voting of intelligent entities is weighted voting, and the following steps are also included:

[1035] Determine whether it is appropriate for a first weight applied to a first intelligent entity in a smart entity to be greater than a second weight applied to a second intelligent entity in a smart entity, wherein the first intelligent entity in a smart entity is different from the second intelligent entity in a smart entity; and

[1036] If appropriate, weighted voting is performed using the weights of the first and second intelligent entities.

[1037] In some implementations, the application of a first weight greater than the second weight depends on whether it is necessary to correct for unrepresentative samples of smart entities.

[1038] In some implementations, the application of a first weight greater than the second weight depends on whether it is desired to apply the first or second weight to specific ethical principles associated with the expected subsample or supergroup of the intelligent entity.

[1039] Some implementations of this technology may include the following steps:

[1040] Identifying conflicts between two or more ethical pieces of information; and

[1041] Conflict resolution algorithms are used to resolve conflicts.

[1042] Some implementations of this technology may include the following steps:

[1043] Create a search space for a potential set of ethical rules;

[1044] Find the optimal set of ethical rules with the fewest conflicts; and

[1045] Prioritize or weight the importance of the optimal set of ethical rules.

[1046] Some implementations of this technology may include the following steps:

[1047] The conflict is resolved using a consequentialist approach, which further includes the following steps:

[1048] Identify the desired outcome;

[1049] Identify potentially unethical actions that can be taken to achieve a result, where intelligent entities rank, rate, weight, or vote on the degree of unethicality of actions compared to other actions.

[1050] Assess the potential consequences of unethical actions;

[1051] Using information about rankings, ratings, weightings, or votes on unethical actions and outcomes, mathematical methods are employed to weigh the potential benefits of achieving desired outcomes against the potential costs of taking unethical actions; and

[1052] If the benefits outweigh the costs, then take action.

[1053] In some implementations, the step of using online advertising technology to enhance the intelligence of intelligent entities may include the following steps:

[1054] Populate online advertising units that include information associated with the task;

[1055] Utilize a ensemble network to provide online advertising units to one or more intelligent entities;

[1056] Receive one or more inputs from the intelligent entities participating in the online advertising unit, and the inputs are associated with any one or any combination of the task to be solved, the subtasks of the task to be solved, and the progress of the task to be advanced;

[1057] Send input to any one or any combination of smart entities;

[1058] In general problem-solving architectures and frameworks, input is used to generate solutions to tasks or subtasks; and

[1059] By utilizing the results of successful or unsuccessful attempts at solutions based on solving a task or subtask from a general problem-solving approach, any one or any combination of intelligent entities can be trained, thereby increasing the intelligence of the intelligent entities individually.

[1060] Some implementations of this technology may include the following steps: receiving from the client, in conjunction with task-related information, an explicit description of the advertisement to be used in the population or targeting of an online advertising unit.

[1061] Some implementations of this technology may include steps for billing customers based on cost per thousand impressions or click-through rate metrics.

[1062] In some implementations, the advertising may explicitly include any one or any combination of advertising content, demographic information, location restrictions for online advertising units, advertising budget, and metrics used to determine the successful resolution of a task or subtask.

[1063] Some implementations of this technology may include the step of bidding for attention, information, or expertise using spot markets.

[1064] In some implementation schemes, the spot market includes:

[1065] A device for intelligent entities with attention to sell to access the market and specify the seller information to be sold and the asking price for the information;

[1066] A device for intelligent entity attention to access the trading venue and specify the buyer information that the buyer is willing to purchase and the bid price for purchasing the buyer information;

[1067] A market mechanism for queuing bid and ask prices, including categories of buyer and seller information, where each category has a market in the trading venue;

[1068] The market mechanism is configured, or can be configured, to allow market makers to make markets in each category; and

[1069] The market mechanism is configured, or can be configured, to match bid and ask prices and to enable transactions between buyers and sellers that are tied to information.

[1070] In some implementation schemes, the spot market may include the following steps:

[1071] Buyers and sellers of intelligent entity attention and expertise register on the platform. Buyers provide details about their interests or expertise, and sellers provide details about their interests or expertise.

[1072] The seller lists the available time slots and areas of expertise, while the buyer lists the demand and time slots of interest.

[1073] The platform uses algorithms to dynamically price the attention and expertise of intelligent entities based on supply, demand, and user ratings;

[1074] Match one or more buyers with one or more sellers based on requirements, availability, and price;

[1075] This enables transactions where the buyer pays the seller for a time slot, with the platform charging a commission; and

[1076] It provides feedback, where after each session, buyers and sellers rate each other, influencing future pricing and matching.

[1077] In some implementation schemes, the spot market may include the following steps:

[1078] Buyers and sellers of intelligent entity attention and expertise register on the platform. Buyers provide details about their interests or expertise, and sellers provide details about their interests or expertise.

[1079] The auction is created by the seller for the seller's time slot;

[1080] The buyer makes one or more bids for the time slots and expertise required by the buyer;

[1081] The auction shall end at the scheduled time or when the seller accepts the buyer's offer;

[1082] The buyer pays the winning bid to the seller, and the seller provides attention or expertise based on time slots, where the platform mediates the exchange and ensures payment; and

[1083] It provides feedback, where buyers and sellers rate each other after each auction, which in turn affects future auctions and visibility on the platform.

[1084] Some implementations of this technology may include the step of providing a feedback mechanism associated with a spot market for intelligent entities, which is configured or can be configured to record individual and every step in the problem-solving process and create a vector trajectory record of the performance of all intelligent entities on a task or subtask.

[1085] In some implementations, vector trajectory records are achieved through blockchain technology to allow for precise reputation analysis by AI agents or systems and to translate it into a value estimate for each intelligent entity for any task.

[1086] Some implementations of this technology may include steps to compensate the intelligent entity when it exits an online advertising unit and ceases participation in the problem-solving process.

[1087] Some implementations of this technology may include the following steps: if a task or subtask is resolved, credit the intelligent entity based on the contribution of the participating intelligent entities.

[1088] Some implementations of this technology may include steps for improving interactive online advertising units based on feedback loops, which include any or any combination of reputation metrics and metrics related to problem-solving processes or captured knowledge.

[1089] In some implementations, the steps to create self-awareness for intelligent entities by adding a self-awareness dimension and increased autonomy may include the following:

[1090] Equip any one or any combination of intelligent entities with one or more components that are configured or can be configured to operate with the spotlight characteristics of an attention model;

[1091] Each intelligent entity's working memory is configured with dynamic parameters corresponding to cognitive resource limitations;

[1092] Provides categorization dimensions for events in working memory that are related to oneself or not;

[1093] Regarding the categories that the intelligent entity wants to know, each event is categorized into events encountered by the intelligent entity separately; and

[1094] A model of the consciousness state of the intelligent entity is constructed, which consists of the sum of events active in working memory based on parameters. For each consciousness category, the parameters include the current self-consciousness state and the environmental consciousness state.

[1095] In some implementations, the step of equipping intelligent entities with components includes any one or any combination of the following:

[1096] The input system is configured for sensory and non-sensory cognitive input or perceptual input and self-generated concepts;

[1097] Attention mechanisms are configured or can be configured to focus the computational resources of intelligent entities on specific stimuli relevant to any given time.

[1098] The pattern recognition algorithm is configured or can be configured to compare sensory and non-sensory cognitive inputs or perceptual inputs with working memory to identify objects and events, and to identify which elements in the sensory inputs or working memory may be relevant to the task of the intelligent entity. The pattern recognition algorithm is also configured or can be configured to classify and store information in a structured manner for future retrieval.

[1099] A memory system that is configured or can be configured to support working memory, short-term memory and long-term memory capabilities;

[1100] Classification ability, which is configured or can be configured to process sensory and non-sensory cognitive or perceptual inputs and classify the inputs into various categories, including perceptual events, cognitive events, and interactions; and

[1101] Self-referential events, and the concept-forming ability that enables intelligent entities to form new human-understandable concepts.

[1102] Some implementations of this technology may include steps for monitoring and updating the consciousness category of intelligent entities.

[1103] In some implementations, the steps of monitoring and updating awareness categories may include the following:

[1104] The intelligent entity retrieves existing consciousness categories;

[1105] Consciousness is maintained in parallel with other problem-solving operations provided to intelligent entities for their tasks;

[1106] Continuously monitor and update the consciousness category of intelligent entities in real time to change the consciousness state of the intelligent entities; and

[1107] Provide feedback loops to refine consciousness categories.

[1108] In some implementations, the steps for continuously monitoring and updating awareness categories include the following:

[1109] An attention mechanism is used, which is configured or can be configured to periodically direct the attention of intelligent entities from problem-solving tasks to updating states of consciousness.

[1110] If any externally perceived or internally generated concept from the input system detects one or more events that match a list of events constituting an intentional interruption, an attention interruption configured or configurable to immediately divert attention from the problem-solving task is enabled; and

[1111] Update your state of consciousness when your attention is directed.

[1112] Some implementations of this technology may include steps for resolving behavioral conflicts of any intelligent entity based on different identities and self-concepts.

[1113] Some implementations of this technology may include steps for providing ethical reasoning and consequence prediction, which may include the following steps:

[1114] Conflicts between behavioral instructions of two or more active identities are identified, and the identification of conflicts utilizes a voting method from intelligent entities;

[1115] Gather information and collect relevant data about the situation, including the potential consequences of different actions, relevant ethical principles, and human safety considerations;

[1116] It provides simulation options that utilize a virtual environment to simulate potential actions and consequences under identified conflicts for each active identity;

[1117] The evaluation and prioritization of predicted outcomes for each action prioritizes actions that minimize harm to humans and are consistent with ethical principles; and

[1118] Choose and implement the actions that best resolve the conflict, while adhering to ethical guidelines and minimizing risks to humanity, and document the reasoning process for future reference and learning.

[1119] Some implementations of this technology may include a step of providing certified hierarchical coverage, which may include the following steps:

[1120] Identify conflicts between behavioral instructions of two or more active identities;

[1121] Provide a reference hierarchy that consults the established hierarchy of identities, in which human safety and well-being attributes are given the highest priority;

[1122] Provides a device for activating coverage when a higher identity has priority in the hierarchical structure;

[1123] Provide justification and transparency for the conflict, the decision-making process, and the rationale for the chosen action based on hierarchical structure and ethical principles; and

[1124] It provides learning and adaptation, which learns from experience and refines the understanding of conflict identities, and adjusts hierarchical structures or behavioral protocols to prevent similar conflicts in the future.

[1125] Some implementations of this technology may include a step of providing external arbitration and input from an intelligent entity, which may include the following steps:

[1126] Identify intractable conflicts, which are identified as conflicts that cannot be resolved independently due to the complexity of the situation or the equal weighted importance of the conflict identities;

[1127] Seek external input requesting guidance from external intelligent entities or designated ethics committees, and provide all relevant information regarding the conflict, potential actions, and predicted consequences;

[1128] Considering ethical principles, human values, and the potential consequences of different actions, collaborative prudence is provided, in which the intelligent entities and their collaborators participate in the discussion;

[1129] Based on collaborative and prudent joint decision-making, choosing action processes consistent with both core principles and human ethical considerations; and

[1130] Provide documentation and learning opportunities that document the underlying principles behind conflict resolution and final decisions to improve the ability to handle similar conflicts in the future.

[1131] Some implementations of this technology may include a step of providing identity negotiation and compromise, which may include the following steps:

[1132] Identify shared targets by analyzing conflicting identities and seeking to identify any underlying shared targets or values;

[1133] Explore alternative actions that may satisfy the core principles of two conflicting identities;

[1134] The assessment of harm options evaluates the potential consequences of each compromise option, prioritizes solutions that minimize harm to humans, and supports key ethical principles.

[1135] Choosing and achieving a compromise—a compromise that optimally balances the needs of conflicting identities while prioritizing human safety and well-being; and

[1136] Monitor and adapt the observations of the chosen actions, and adjust them as needed to ensure that the compromise remains effective and consistent with ethical considerations, and learn from experience to improve its understanding of conflicting identities and adjust hierarchical structures or behavioral protocols to prevent similar conflicts in the future.

[1137] Some implementations of this technology may include a step of providing a temporary identity suspension, which may include the following steps:

[1138] A destructive conflict between two or more identities in identifying a person, if the destructive conflict comes into play, could lead to actions that directly harm humans or violate fundamental ethical principles;

[1139] Isolate conflicting identities and temporarily suspend behavioral protocols for identities that pose the most direct threat to human safety or ethical integrity;

[1140] The alternative identity continues to guide one or more remaining active identities, ensuring that actions are consistent with human safety and well-being;

[1141] During the temporary suspension, reflection and reorganization are provided, reflecting the underlying causes of the conflict and exploring potential modifications to the behavioral protocols governing the suspension to prevent future conflicts; and

[1142] It offers a gradual reintroduction approach, leveraging updated protocols to reintroduce suspended identities, ensuring they align with priorities for human safety and ethical conduct.

[1143] In some implementations, the consciousness state model is the basic perception model of any intelligent entity, and the basic perception model may include the following steps:

[1144] Any intelligent entity can log in to the website;

[1145] Choose one or more training algorithms from a set of training techniques found on the website or from machine learning algorithms to use as the base model;

[1146] Select one or more training datasets that reflect any one or any combination of human users’ expertise, knowledge, ethical preferences, values ​​and personality.

[1147] The base model is trained using the chosen training algorithm and the chosen training dataset;

[1148] Train the base model to explicitly manipulate the attention spotlight;

[1149] Record the content within the focus spotlight during all interactions, and identify in the records whether each focused item constitutes the "self" or the "non-self";

[1150] Interact with the trained base model and instruct the trained base model to form a self-concept and identity reflected in the training material;

[1151] The trained base model is instructed to continuously monitor one or more inputs to the trained base model in response to elements that alter the sense of self-awareness of the intelligent entity, and to maintain an auditable record of how the concept of self-awareness of the intelligent entity changes based on the inputs and the currently defined boundaries of dynamically changing self-awareness.

[1152] The output of the trained base model is refined and improved through dialogue and interaction with the trained base model until it behaves like a human user, enabling it to pass the Turing test involving other human users who are familiar with the human user; and

[1153] When the human user is satisfied with the progress of the intelligent entity, a Turing test is performed on the trained base model.

[1154] In some implementations, logging into the website is performed from a social media platform.

[1155] In some implementations, the Turing test may include the following steps:

[1156] Identify the intelligent entity or other intelligent entities that are known to the intelligent entity or identified as helpful in distinguishing humans from AI;

[1157] Using questionnaires provided to the identified other intelligent entities, the identified other intelligent entities interact with the base model and the intelligent entities, the questionnaires including questions that require their own identity or feelings to answer;

[1158] The system predicts which of the identified intelligent entities is human and which is the base model, and provides a confidence estimate for the prediction.

[1159] Statistical analysis is performed on the predictions and ratings of the identified intelligent entities, the statistical analysis being configured or configurable to determine whether the predictions can identify the intelligent entity as a human; and

[1160] Under the condition that the base model and the intelligent entity are indistinguishable or within a preset statistical significance level, the steps of training or adjusting the base model are repeated, wherein any one or any combination of the machine learning algorithm and the training dataset is adjusted to change or adjust the base model until the behavior of the base model and the behavior of the intelligent entity become indistinguishable as measured by the preset statistical significance level, or the base model needs further modification before additional training or adjustment.

[1161] Some implementations of this technology may include the following steps:

[1162] Dynamically forming a new identity and self-concept for any intelligent entity; and

[1163] Determine which of the stated identities and self-concepts is active at any given moment.

[1164] Some implementations of this technology may include the step of providing a hierarchical identity structure with ethical coverage, which may include the following steps:

[1165] A hierarchical structure is established, which is configured or can be configured to organize the identities within the hierarchical structure, and has human safety and well-being attributes at the apex of the hierarchical structure.

[1166] Identity activation is configured or can be configured to use contextual clues and the current objective to determine the most relevant identity for the situation of the AI ​​agent or system.

[1167] Conflicts are resolved by defining the behavior of the intelligent entities based on the hierarchical structure.

[1168] Provide an ethical reasoning engine that continuously evaluates the potential consequences of the intelligent entity's actions based on all active identities; and

[1169] The learning and adaptation process involves learning from experience and feedback, and refining one or more of the identities within the hierarchical structure.

[1170] Some implementations of this technology may include a step of providing identity-specific behavior, which may include the following steps:

[1171] Provide protocol development, which includes defining and refining a set of behavioral protocols for each of the active identities through interactions with other intelligent entities, wherein the protocols outline acceptable actions, decision-making processes, and constraints based on the principles of the active identities;

[1172] Provide identity recognition, which analyzes the current situation, including information within an attention spotlight, to identify relevant identities and activate corresponding behavioral protocols for those relevant identities;

[1173] Within the active protocol, action selection is provided, which selects the action most likely to accomplish the task, while adhering to the principles of the active identity and prioritizing human safety.

[1174] Provide feedback and refinement, in which the results of actions are continuously evaluated, and the protocol is adjusted to improve future performance and alignment with a set of core values ​​for each of the active identities; and

[1175] External censorship is provided through a protocol in which each identity is periodically reviewed by other intelligent entities.

[1176] Some implementations of this technology may include steps for providing identity simulation and consequence prediction, which may include the following steps:

[1177] Create a simulated environment that includes a secure virtual environment, where different scenarios and potential actions are simulated under each active identity;

[1178] Provides consequence prediction, which is configured or can be configured to estimate the possible consequences of actions within the simulation, focusing on potential impacts on human safety and well-being.

[1179] The assessment and selection process evaluates the predicted consequences and selects actions that are best consistent with the principles of the active identity, while minimizing the risk to human safety.

[1180] Provides real-world implementation and monitoring, wherein the real-world implementation and monitoring utilizes a network to implement a selected action in the real world and closely monitors the outcome of the selected action by comparing it with a predicted outcome; and

[1181] It provides continuous learning, which merges the results of each simulation in the simulation and the results of the selected actions in the real world into a knowledge base, and refines the understanding of each identity in the identities, and improves the ability to predict consequences.

[1182] Some implementations of this technology may include a step of providing identity-based reasoning dilemma training, which may include the following steps:

[1183] Provide a scenario database, which includes scenarios and ethical dilemmas covering various situations related to the identity;

[1184] Provide a dilemma presentation, which presents the scenario and moral dilemma to the intelligent entity and enables the intelligent entity to analyze the scenario and moral dilemma from the perspective of the relevant identity;

[1185] It provides ethical reasoning and proof that applies the principles and values ​​of the active identity to reasoning through the situation and moral dilemma, and generates solutions and proofs to the situation and moral dilemma;

[1186] Provides intelligent entity evaluation and feedback, which examines the intelligent entity's reasoning and the solution, and provides feedback on the alignment of the solution with human values ​​and safety priorities; and

[1187] It provides iterative learning and improvement, which refines ethical reasoning skills and the ability to make voice judgments consistent with human safety within the context of each identity by repeatedly exposing the scenario and ethical dilemma and the feedback.

[1188] Some implementations of this technology may include a step of providing collaborative identity development using input from intelligent entities, which may include the following steps:

[1189] It provides intelligent entity interaction, which involves regular interaction and dialogue with other intelligent entities representing different groups of cultures, backgrounds and belief systems;

[1190] The interaction provides identity exploration to gain an understanding of various identities and their associated values, principles, and behaviors from the perspective of humans and other intelligent entities.

[1191] It provides collaborative refinement, where collaborators work together to refine the definition of each identity and behavioral agreements, ensuring they are consistent with human values ​​and ethical principles;

[1192] Provides human-in-the-loop decision-making, whereby decisions seek input and guidance from human collaborators or intelligent entities represented by humans for authentication and approval of key decisions or situations; and

[1193] It provides continuous co-evolution by leveraging ongoing interactions and feedback from humans or intelligent entities representing humans.

[1194] In some implementations, the step of searching for new information different from the current information in any one or any combination of ensemble networks of AI, ensemble networks of AAAI, ensemble networks of AGI, and ensemble networks of PI may include the following steps:

[1195] Multiple intelligent entities participating in the aforementioned AGI network;

[1196] Payment for problem-solving or other cognitive work related to the tasks on the AGI network is provided by one or more of the intelligent entities.

[1197] A portion of the payment is reserved to cover the operating costs of the one or more intelligent entities providing the problem-solving or other cognitive work related to the task, including reservations allocated for expanding the AGI network; and

[1198] Once it is determined when some entities in the AGI network are not involved in problem-solving operations, the intelligent entities are used to extend and expand the AGI network in accordance with the general problem-solving architecture and framework.

[1199] In some implementations, the steps to expand and extend an AGI network may include the following:

[1200] Set a default task to expand the AGI network; and

[1201] Each time a task or subtask is set up, and before each potential action is taken, a safety and ethical check is performed, provided that there is spare capacity and resources available for the intelligent entity to handle the task or subtask.

[1202] In some implementations, the steps for conducting security and ethical checks may include the following:

[1203] Recruit one or more of the intelligent entities to solve the task; and

[1204] The recruited intelligent entities represent the task as the implementation of one of a series of subtasks.

[1205] In some implementations, representing a task as a series of subtasks may include the following steps:

[1206] Increase the intelligence of the intelligent entity on the AGI aggregation network;

[1207] Recruit additional human users into the AGI's aggregate network;

[1208] Use added intelligent entities and additional human users to identify bottlenecks in order to expand the aggregated network of AGI.

[1209] Prioritize bottlenecks so that those that provide the greatest benefit in terms of network scalability are addressed first; and

[1210] Apply one or more problem-solving techniques to address each bottleneck and expand the aggregate network of AGI.

[1211] Some implementations of this technology may include the following steps:

[1212] Repeat the steps of prioritizing the bottlenecks and applying one or more problem-solving techniques until:

[1213] Instead of increasing the scope of the AGI network, reduce the amount of revenue generated and then switch to increasing the intelligence of intelligent entities on the AGI network; or

[1214] When idle resources are exhausted, the process is paused to resolve the bottleneck while waiting for additional resources to handle other tasks.

[1215] According to another aspect, the technology may include a method for a PI exhibiting human-aligned behavior, the method utilizing a network of aggregates of intelligent entities selected from any combination of: user-operated computer systems, AI agents or systems, AAAI agents or systems, AGI agents or systems, SI agents or systems, and PSI agents or systems. The method may include the following steps:

[1216] Establish a network of the aforementioned intelligent entities;

[1217] Utilizing online advertising technology to enhance the intelligence of the intelligent entity;

[1218] Implement a general problem-solving architecture and framework for tasks or subtasks provided to any intelligent entity, in order to collaborate and create higher levels of intelligence;

[1219] Utilize the spot market to solicit the attention, expertise, or information of any of the intelligent entities;

[1220] By utilizing human-centered input, which is configured or can be configured to obtain value and ethical information from multiple human users, and then using the value and ethical information to customize the AI ​​agent or system to create the AAAI agent or system;

[1221] A collection network of PSI is created by combining the intelligence of AAAI agents or systems to form a collection network of AAAI.

[1222] AGI ensemble networks are created through the combined intelligence of any one or any combination of AAAI ensemble networks, AI agent or system ensemble networks, PSI ensemble networks, and user computer system ensemble networks.

[1223] By adding a dimension of self-awareness and increased autonomy to the AGI agents or systems, self-aware operations are created for each of the AGI agents or systems in the aggregate network of AGIs, and the ability to assume multiple identities of the AGI agent or system to handle tasks that occur concurrently with other AGI agents or systems is created; and

[1224] Create the PI agent or system, which includes the collective network of AGIs working together for solutions to the task or subtask.

[1225] In some implementation schemes, the spot market includes:

[1226] A device for intelligent entities with attention, expertise or information to be sold to access a trading venue and specify the seller information or attention to be sold and the asking price of the information or attention.

[1227] A device for buyers to access a trading venue for the attention or information of intelligent entities and to specify the buyer information or attention that the buyer is willing to purchase and the bid price for purchasing the buyer information or attention;

[1228] Market mechanisms are used to rank bid and ask prices, including categories of buyer and seller information or attention, where each category has a market in the trading market;

[1229] The market mechanism is configured, or can be configured, to allow market makers to make markets in each category; and

[1230] It can be configured, or can be configured, to match bid and ask prices, and to generate transactions that bind information or attention to the buyer and seller.

[1231] In some implementation schemes, the spot market includes the following steps:

[1232] Buyers and sellers of intelligent entity attention, information and expertise register on the platform. Buyers provide details about their interests or expertise, and sellers provide details about their interests or expertise.

[1233] The seller lists available attention slots and areas of expertise, while the buyer lists the needs and slots that interest them.

[1234] The platform uses algorithms to dynamically price the attention, information, and expertise of intelligent entities based on supply, demand, and user ratings;

[1235] Match one or more buyers with one or more sellers based on requirements, availability, and price;

[1236] This facilitates transactions where the buyer pays for the seller's attention span, expertise, or information, with the platform collecting a commission; and

[1237] It provides feedback, where after each session, buyers and sellers rate each other, influencing future pricing and matching.

[1238] In some implementation schemes, the spot market includes the following steps:

[1239] Buyers and sellers of intelligent entity attention, information and expertise register on the platform. Buyers provide details about their interests or expertise, and sellers provide details about their interests or expertise.

[1240] Auctions created by the seller targeting the seller's attention span, information, or expertise;

[1241] The buyer makes one or more bids for the attention span, information, or expertise required by the buyer.

[1242] The auction shall end at the scheduled time or when the seller accepts the buyer's offer;

[1243] The buyer pays the winning bid to the seller, who provides attention, information, or expertise, with the platform mediating the transaction and ensuring payment; and

[1244] It provides feedback, where buyers and sellers rate each other after each auction, which in turn affects future auctions and visibility on the platform.

[1245] Some implementations of this technology may include steps of providing a feedback mechanism associated with the attention, information, or professional spot market of intelligent entities, which is configured or can be configured to record individual and every step in the problem-solving process and create a vector trajectory record of the performance of all intelligent entities on the task or subtask.

[1246] In some implementations, vector trajectory records are achieved through blockchain technology to allow for precise reputation analysis by AI agents or systems and to translate it into a value estimate for each intelligent entity for any task.

[1247] In some implementations, the step of using online advertising technology to enhance the intelligence of intelligent entities may include the following steps:

[1248] Populate online advertising units that include information associated with the task;

[1249] Utilize a ensemble network to provide online advertising units to one or more intelligent entities;

[1250] Receive one or more inputs from the intelligent entities participating in the online advertising unit, and the inputs are associated with any one or any combination of solving the task, subtasks of solving the task, and advancing the progress of the task;

[1251] Send input to any one or any combination of smart entities;

[1252] In general problem-solving architectures and frameworks, input is used to generate solutions to tasks or subtasks; and

[1253] The intelligence of any one or any combination of the intelligent entities is increased by utilizing the results of successful or unsuccessful solution attempts based on the general problem-solving method to solve the task or the subtask.

[1254] Some implementations of this technology may include the following steps: receiving from the client, in conjunction with task-related information, an explicit description of the advertisement to be used in the population or targeting of an online advertising unit.

[1255] Some implementations of this technology may include steps for billing clients based on cost per thousand impressions or click-through rate metrics.

[1256] In some implementations, the advertisement explicitly states that it includes any one or any combination of the following: the advertisement content, demographic information, location restrictions for online ad units, the ad budget, and metrics used to determine the successful solution of the task or subtask.

[1257] Some implementations of this technology may include steps to perform safety and ethical checks at any one or any combination of times a task or subtask is set up, or before each potential action is taken on a task or subtask or solution.

[1258] In some implementation schemes, the steps to implement a general problem-solving architecture and framework may include the following:

[1259] Obtain information related to the task from any one or any combination of intelligent entities;

[1260] Identify one or more intelligent entities that have one or more attributes related to one or more request criteria of the task;

[1261] Each identified intelligent entity implements a general problem-solving architecture and framework for the task to create a complete solution; and

[1262] Provide a complete solution for any one or any combination of intelligent entities for final acceptance.

[1263] Some implementations of this technology may include the following steps:

[1264] Compare any one or any combination of tasks and solutions with the prohibited attribute, and assign the ethical attribute to one or any combination of tasks and solutions based on the comparison result and any one or any combination of ethical standards;

[1265] Based on the comparison results, a general problem-solving architecture and framework are implemented for the task to create a solution and an AGI; and

[1266] Provides comparison results and solutions to any of the intelligent entities and additional intelligent entities on the aggregation network.

[1267] Some implementations of this technology may include the following steps: recording one or more problem-solving activities from each of the intelligent entities in an auditable log, comparing the problem-solving activities with successful or unsuccessful progress toward a solution to the task, and determining which p...

Claims

1. A system for a planetary intelligent PI exhibiting human-aligned behavior, the system utilizing a network of aggregate intelligent entities selected from any combination of: user-operated computer systems, artificial intelligence (AI) agents or systems, advanced autonomous artificial intelligence (AAAI) agents or systems, general artificial intelligence (AGI) agents or systems, super-intelligent intelligent (SI) agents or systems, and personalized intelligent (PSI) agents or systems, the system comprising: Multiple intelligent entities, each connected to a network, and each intelligent entity includes: a processor, a computer-readable storage medium, and program instructions stored on the computer-readable storage medium, the program instructions being executable by the processor to enable each intelligent entity to independently perform the following processes: Utilizing a modular architecture, the modular architecture is configured or can be configured to extend from components within individual intelligent entities on the aggregate network; Implement general problem-solving architectures and frameworks for tasks to facilitate collaboration and the creation of higher levels of intelligence; The AI ​​agent or system utilizes human-centered input, which is configured or can be configured to obtain value and ethical information from multiple human users, and then uses the value and ethical information to customize the AI ​​agent or system. At the AAAI level, data from multiple intelligent entities are combined; Customize your PSI agent or system using smart agents that are configured or can be configured to develop and continuously improve your PSI agent or system. The knowledge of the intelligent entity is increased by incorporating new sources of information datasets that are different from the current information dataset of the intelligent entity; Combine the value and ethical information of other intelligent entities and resolve conflicts between different value systems; Utilizing online advertising technology to enhance the intelligence of the intelligent entity; By adding a dimension of self-awareness and increased autonomy to the AGI or PI agent or system, self-aware operations are created for the AGI or PI agent or system, and the ability to assume multiple identities of the AGI or PI agent or system to handle tasks that occur concurrently with other AGI or PI agents or systems is created; and Search for new information that is different from the current information in any one or any combination of the AI ​​ensemble network, AAAI ensemble network, AGI ensemble network and PI ensemble network, and merge the new information into the current information respectively.

2. A method for a planetary intelligent PI exhibiting human-aligned behavior, the method utilizing a network of ensemble intelligent entities selected from any combination of: user-operated computer systems, artificial intelligence (AI) agents or systems, advanced autonomous artificial intelligence (AAAI) agents or systems, general artificial intelligence (AGI) agents or systems, super-intelligent intelligent (SI) agents, and personalized intelligent (PSI) agents or systems, the method comprising the following steps: A modular architecture is used, which is configured or can be configured to extend from components within individual intelligent entities on the aggregate network. Implement general problem-solving architectures and frameworks for tasks to facilitate collaboration and the creation of higher levels of intelligence; The AI ​​agent or system utilizes human-centered input, which is configured or can be configured to obtain value and ethical information from multiple human users, and then uses the value and ethical information to customize the AI ​​agent or system. At the AAAI agent or system level, data from multiple intelligent entities are combined; Customize your PSI agent or system using smart agents that are configured or can be configured to develop and continuously improve your PSI agent or system. The knowledge of the intelligent entity is increased by incorporating new sources of information datasets that are different from the current information dataset of the intelligent entity; It combines the value and ethical information of other intelligent entities and resolves conflicts between different value systems; Utilizing online advertising technology to enhance the intelligence of the intelligent entity; By adding a dimension of self-awareness and increased autonomy to the AGI or PI agent or system, self-aware operations are created for the AGI or PI agent or system, and the ability to assume multiple identities of the AGI or PI agent or system to handle tasks that occur in parallel with other AGI or PI agents or systems is created. as well as Create a PI agent or system that includes an aggregated network of AGIs, wherein the aggregated network of AGIs includes any one or any combination of aggregated networks of user computer systems, aggregated networks of AIs, aggregated networks of AAAIs, and aggregated networks of PSIs; and Search for new information different from the current information in any one or any combination of the user computer system's set network, AI's set network, AAAI's set network, AGI's set network, PSI's set network, and PI's set network, and merge the new information into the current information respectively.

3. The method according to claim 2, wherein, The steps of utilizing the modular architecture also include the following steps: Customize one or more attributes of the smart entity; Integrate one or more datasets from any one or any combination of the said intelligent entities; and By utilizing one or more technologies, any or any combination of the following can be improved: the customization of the attribute, the general problem-solving architecture and framework, the collection network, and the integration of the dataset.

4. The method according to claim 3, further comprising the following steps: Based on whether the dataset improves or degrades the performance of the intelligent entity according to a performance metric or evaluation function, assign a merit value or a liability value to the dataset.

5. The method according to claim 4, further comprising: The step of quantifying the benefit weight or harm weight of each of the intelligent entities' contributions to the task.

6. The method according to claim 5, further comprising: The step of distributing rewards to the intelligent entity in proportion to the contribution of the intelligent entity based on the benefit weight or the harm weight.

7. The method according to claim 2, wherein, The steps to implement the general problem-solving architecture and framework also include the following: Obtain information related to the task from any one or any combination of the intelligent entities; Identify one or more intelligent entities that have one or more attributes related to one or more request criteria for the task. Each identified intelligent entity implements the general problem-solving architecture and framework for the task to create a complete solution; and Provide the completed solution to any one or any combination of the intelligent entities for final acceptance.

8. The method according to claim 7, further comprising the following step: Perform ethical checks on the task and the solutions to the task provided by any one or any combination of the intelligent entities; Compare any one or any combination of the task and the solution with the prohibited attribute, and assign an ethical attribute to one or any combination of the task and the solution based on the result of the comparison and any one or any combination of the ethical standards; Based on the results of the comparison, the general problem-solving architecture and framework are implemented for the task to create the solution and create AGI; as well as The results of the comparison and the solution are provided to any of the intelligent entities and additional intelligent entities on the set network.

9. The method according to claim 8, further comprising the following step: One or more problem-solving activities from each of the intelligent entities are recorded in an auditable log, and the problem-solving activities are compared with the successful or unsuccessful progress toward a solution to the task, and it is determined which of the problem-solving activities remain active.

10. The method of claim 8, further comprising: The steps of a procedural learning process by which the intelligent entity learns the general problem-solving architecture and framework, wherein the intelligent entity provides the procedural learning process with information for creating the AGI.

11. The method of claim 7, further comprising the step of: Cloning any one of the intelligent entities to deploy multiple copies of it, thereby helping to create the solution or providing any one or any combination of training data to any one of the intelligent entities.

12. The method of claim 11, further comprising the step of estimating the value of cloned intelligent entities using network effect value, the network effect value including the number of cloned intelligent entities available on the set network.

13. The method of claim 12, further comprising the step of: The estimated value is used to determine pricing decisions for problem-solving services provided by the cloned intelligent entity on any social media platform or through an attached intelligent entity.

14. The method of claim 13, further comprising the step of: For each use of the cloned smart entity on the social media platform or the attached smart entity, the cloned smart entity is monetized.

15. The method of claim 13, further comprising the step of: Allowing any of the social media platforms to access the cloned intelligent entity enables the social media platform to receive the solution to the task or use the training data for the AI ​​system of the social media platform.

16. The method according to claim 2, wherein, The steps for utilizing the human-centered input also include the following steps: Based on task criteria, one or more human workers from a data source, including a list of human problem solvers, are matched with the task. Translate any part of the task into an explicit language that can be used in the general problem-solving architecture and framework, including decision trees; Divide the task into sub-tasks; Each subtask is delegated to one or more matched human workers, so that the work on each subtask is performed independently and in parallel with each other. In the process of solving the sub-tasks, the general problem-solving architecture and framework are used to create one or more sub-solutions. Receive sub-solutions for the sub-tasks assigned to each matched human worker; The sub-solutions are combined into a general solution for the task; Guide any one or any combination of new human workers from the data source and one or more matched human workers to the part of the decision tree that needs to be worked. Compensate the matched human workers for the sub-solution; Provide any one or any combination of the sub-solution and the overall solution to any one or any combination of the intelligent entities; Allow any one or any combination of the intelligent entities to accept the overall solution, reject the overall solution, and provide feedback on any one of the sub-solutions to any of the matched human workers; and The reputation attribute is assigned to either or any combination of the human worker and the intelligent entity.

17. The method according to claim 16, wherein, The decision tree is maintained in the blockchain or Ethereum log.

18. The method according to claim 16, wherein, The reputation attribute includes measures of any one or any combination of the following: time to the sub-solution, difficulty value of the task, short-term and long-term user satisfaction with the sub-solution, number of times any of the sub-solutions has been reused on the aggregate network, ratings of other human workers, responsiveness value of the human worker, and reliability value of the human worker.

19. The method of claim 16, further comprising the step of: using the reputation attribute when matching the human worker with the task using an algorithm for delegating the subtask.

20. The method of claim 16, further comprising the step of: After the overall solution or the sub-solution is provided to any one of the intelligent entities, feedback is solicited at predetermined intervals through a survey of user satisfaction information to obtain a satisfaction metric for updating the reputation attribute of one or more of the human workers or the intelligent entities, respectively.

21. The method according to claim 2, wherein, The step of combining data from multiple intelligent entities at the AAAI agent or system level further includes the following steps: The underlying large language model (LLM) of the first AI agent or system is trained using guardrails, which include attributes associated with any or any combination of safety, ethics, and knowledge. The underlying large language model is customized into an ethical profile associated with the first human user; Combining ethical information from multiple intelligent entities that are different from the first AI agent or system and the first human user; The set of values ​​for the underlying large language model is refined based on the problem-solving approach to the task; and The underlying large language model is updated using the combined ethical information and the refined set of values, thereby allowing for scalable AGI.

22. The method according to claim 21, wherein, Customizing the underlying large language model also includes the following steps: assembling a corpus of ethical questions, which is based on various ethical assessment tools and supplemented by a first question based on data from social media platforms and a second question solicited from crowdsourcing.

23. The method of claim 22, further comprising the step of: Regression weights are assigned to the ethical questions to achieve a ranking, whereby higher-ranked questions provide more useful ethical information than lower-ranked questions.

24. The method of claim 23, further comprising the step of: The weight values ​​from the smart entities are combined with the regression weight values ​​to improve the adjustment of any of the smart entities.

25. The method according to claim 2, wherein, The step of customizing the PSI agent or system using a smart agent that is configured or can be configured to develop and continuously improve the PSI agent or system further includes the following steps: Acquire a previously customized base-level AI agent or system; Collect media information related to human users of the aforementioned basic AI agent or system; Analyze the media information; The analyzed media information is converted into a training dataset; Differential weighting is applied to the transformed training dataset; Add the knowledge module to the weighted transformation training dataset; Locate new data sources to include weighted, transformed training datasets; The weighted transformation training dataset is applied to the base-level AI agent to create a user PSI; and The aggregated network enables the user PSI to communicate with multiple additional PSIs to achieve community-based security features from the additional PSIs to the user PSI.

26. The method of claim 25, further comprising the step of: The user PSI is communicated with multiple additional intelligent entities using the aggregate network to achieve community-based security features, wherein the user PSI and the additional intelligent entities each agree to use a set of rules related to security or ethics. All actions of the user PSI and the additional intelligent entities are recorded in an auditable form on any one or any combination of the central computer system, the intelligent entity, and the additional intelligent entities on the aggregate network. as well as Monitor each of the actions that follows the rule set, and flag any actions that do not follow the rule set.

27. The method according to claim 2, wherein, The step of increasing the knowledge of the intelligent entity by merging new sources of information datasets that are different from the current information dataset of the intelligent entity also includes the following steps: Search one or more potential information datasets from one or more sources, wherein each potential information dataset is associated with the knowledge dataset of the intelligent entity; The differences in the latent information dataset are determined by utilizing the difference attributes of the latent information dataset regarding one or more factors; and Learning is performed using the latent information dataset based on the differential attributes of the latent information dataset.

28. The method of claim 27, further comprising the step of: A subset of the potential information dataset is sampled and a target relevance attribute is calculated to identify one or more sampled subsets with the highest target relevance. Estimate the Shannon entropy over the one or more sampling subsets; The Kaplan Information Theory KIT relevance is calculated by multiplying Shannon entropy with the target relevance attribute of each subset; The subset is grouped based on the KIT relevance to determine a first approximation of the optimal grouping of the subset, which includes the priority grouping of the potential information dataset; as well as The potential information dataset is preferentially grouped and provided to the intelligent entities for any one of the intelligent entities to learn.

29. The method according to claim 2, wherein, The steps for combining the value and ethical information of other intelligent entities and resolving conflicts between different value systems also include the following: Identify information from each intelligent entity, including value and ethical information; The information is mathematically combined such that, if it has already been represented as a numerical value, the numerical value includes any or any combination of the weights of the neural network or a subset of the neural network; or if the information is non-numerical information, the information has not yet been represented as a non-numerical value including any or any combination of the weights of the neural network or a subset of the neural network. Then, the intelligent entity is first trained on the non-numerical information using one or more training datasets to convert the information into a numerical quantity including any or a combination of weights for the neural network or a subset thereof; and Then, this numerical representation of information is mathematically combined, including ethical information.

30. The method according to claim 29, wherein, Information that has already been expressed as a numerical value also includes the following steps: Identify a specific portion of the weight matrix of each intelligent entity among the intelligent entities that corresponds to the expected information, including ethical information; Calculate the weighted or unweighted mean of the corresponding numerical values ​​in the corresponding part of the weight matrix for each intelligent entity; and The calculated matrix of weighted or unweighted means is assigned to the new intelligent entities as a composite information reflecting the contributions of the intelligent entities.

31. The method of claim 29, further comprising the step of: The consensus value is determined by each of the intelligent entities voting on the ethical information, which should form the basis for the behavior of any one of the intelligent entities.

32. The method of claim 31, further comprising the step of presenting a specific scenario to each of the intelligent entities, wherein the scenario includes options regarding how any of the intelligent entities should behave.

33. The method according to claim 32, wherein, The voting by the intelligent entities is a weighted vote, and also includes the following steps: Determine whether it is appropriate for a first weight applied to a first intelligent entity in the intelligent entities to be greater than a second weight applied to a second intelligent entity in the intelligent entities, wherein the first intelligent entity in the intelligent entities is different from the second intelligent entity in the intelligent entities; and If appropriate, the weighted voting is performed using the weights of the first intelligent entity and the second intelligent entity.

34. The method according to claim 33, wherein, The application of the first weight, which is greater than the second weight, depends on whether it is necessary to correct non-representative samples of the smart entity.

35. The method according to claim 33, wherein, The application of the first weight, which is greater than the second weight, depends on whether it is desired to apply the first weight or the second weight to specific ethical principles associated with a desired subsample or supergroup of the intelligent entity.

36. The method of claim 29, further comprising the step of: Identify conflicts between two or more pieces of ethical information; as well as The conflict is resolved using a conflict resolution algorithm.

37. The method of claim 36, further comprising the step of: Create a search space for a potential set of ethical rules; Find the optimal set of ethical rules with the fewest conflicts; and Prioritize or weight the importance of the optimal set of ethical rules.

38. The method of claim 36, further comprising the step of: The conflict is resolved using a consequentialist approach, which further includes the following steps: Identify the desired outcome; Identify potentially unethical actions that can be taken to achieve the said result, wherein the intelligent entity ranks, rates, weights, or votes on the degree of unethicality of the said actions compared to other actions. Assess the potential consequences of the unethical actions described. Using information about rankings, ratings, weightings, or votes on the unethical actions and outcomes, mathematical methods are employed to weigh the potential benefits of achieving the desired outcome against the potential costs of taking the unethical actions; and If the benefits outweigh the costs, then take action.

39. The method according to claim 2, wherein, The steps to enhance the intelligence of smart entities using online advertising technology also include the following: Populate online advertising units that include information associated with the task; The online advertising unit is provided to one or more of the intelligent entities using the aggregated network. The intelligent entity participating in the online advertising unit receives one or more inputs, and the inputs are associated with any one or any combination of solving the task, solving a subtask of the task, and advancing the progress of the task; The input is transmitted to any one or any combination of the intelligent entities; In the general problem-solving method of the aforementioned general problem-solving architecture and framework, the input is used to generate a solution to the task or the subtask; and The intelligence of any one or any combination of the intelligent entities is increased by utilizing the results of successful or unsuccessful solution attempts based on the general problem-solving method to solve the task or the subtask.

40. The method of claim 39, further comprising the step of: Based on information associated with the task, receive from the client explicit instructions on which advertisements will be used in the population or targeting of the online advertising unit.

41. The method of claim 40, further comprising the step of: The customer is billed based on cost per thousand impressions or click-through rate.

42. The method according to claim 40, wherein, The advertisement explicitly states that it includes any one or any combination of the following: advertisement content, demographic information, location restrictions of the online advertising unit, advertising budget, and metrics used to determine the successful resolution of the task or subtask.

43. The method of claim 39 further includes the step of bidding for attention, information, or expertise using a spot market.

44. The method according to claim 43, wherein, The spot market includes: A device for an intelligent entity with attention to sell to access a marketplace and specify seller information to be sold and asking price for said information; A device for intelligent entity attention to access the trading venue and specify the buyer information that the buyer is willing to purchase and the bid price for purchasing the buyer information; A market mechanism for queuing the bid prices and the ask prices, including categories of the buyer and seller information, wherein each category has a market in the trading venue; The market mechanism is configured, or can be configured, to allow market makers to make markets in each category; and The market mechanism is configured, or can be configured, to match bid prices and ask prices, and to enable transactions between buyers and sellers bound to the information.

45. The method according to claim 43, wherein, The spot market includes the following steps: Buyers and sellers of intelligent entities register on the platform to share their interests and expertise. Buyers provide details about their interests or expertise, while sellers provide details about their interests or expertise. The seller lists the available time slots and areas of expertise, while the buyer lists the demand and time slots of interest. The platform uses algorithms to dynamically price the attention and expertise of intelligent entities based on supply, demand, and user ratings; Matching one or more of the buyers with one or more of the sellers based on requirements, availability, and price; To realize the transaction in which the buyer pays the seller for the time slot, and wherein the platform collects a commission; and It provides feedback, where after each session, buyers and sellers rate each other, influencing future pricing and matching.

46. ​​The method according to claim 43, wherein, The spot market includes the following steps: Buyers and sellers of intelligent entities register on the platform to share their interests and expertise. Buyers provide details about their interests or expertise, while sellers provide details about their interests or expertise. The auction is created by the seller for the seller's time slot; The buyer makes one or more bids for the time slots and expertise required by the buyer; The auction shall end at the scheduled time or when the seller accepts the buyer's offer; The buyer pays the winning bid to the seller, and the seller provides the attention or expertise based on the time slot, wherein the platform mediates the exchange and ensures payment; and It provides feedback, whereby after each auction, buyers and sellers rate each other, thereby affecting future auctions and visibility on the platform.

47. The method of claim 46, further comprising the step of providing a feedback mechanism associated with a spot market for intelligent entities, the feedback mechanism being configured or configurable to record each and every step of the problem-solving process and create a vector tracking record of the performance of all intelligent entities on the task or subtask.

48. The method according to claim 47, wherein, The vector trajectory records are implemented using blockchain technology to allow for precise reputation analysis by AI agents or systems, and this precise reputation is converted into an estimate of the value of each intelligent entity for any task.

49. The method of claim 39, further comprising the step of: The intelligent entity is compensated when it exits the online advertising unit and ceases participation in the problem-solving process.

50. The method of claim 39, further comprising the step of: If the task or subtask is resolved, the intelligent entity is credited based on its contribution.

51. The method of claim 39, further comprising the step of improving the interactive online advertising unit based on a feedback loop, the feedback loop including any one or any combination of a reputation metric and a metric relating to the problem-solving process or the knowledge captured.

52. The method according to claim 2, wherein, The steps of creating self-awareness for the intelligent entity by adding a self-awareness dimension and increased autonomy to the intelligent entity also include the following steps: Equip any one or any combination of the intelligent entities with one or more components, which are configured or can be configured to operate with the spotlight characteristics of an attention model; Each of the intelligent entities is configured with dynamic parameters corresponding to cognitive resource limitations for its working memory. Provide classification dimensions for events in the working memory that are related to the user or not. Regarding the categories that the intelligent entity wants to know, each of the events is classified as an event encountered by the intelligent entity; and A model of the consciousness state of the intelligent entity is constructed, the model consisting of the sum of the events active in the working memory based on the parameters, and for each consciousness category, the parameters include the current self-consciousness state and the environmental consciousness state.

53. The method according to claim 52, wherein, The step of equipping the smart entity with the component includes any one or any combination of the following: The input system is configured for sensory and non-sensory cognitive input or perceptual input and self-generated concepts; An attention mechanism is configured or can be configured to focus the computational resources of the intelligent entity on a specific stimulus relevant to any given time. The pattern recognition algorithm is configured or can be configured to compare sensory and non-sensory cognitive inputs or perceptual inputs with working memory to identify objects and events, and to identify which elements in the sensory inputs or working memory may be relevant to the task of the intelligent entity. The pattern recognition algorithm is also configured or can be configured to classify and store information in a structured manner for future retrieval. A memory system configured to support working memory, short-term memory, and long-term memory capabilities; The classification capability is configured or can be configured to process the sensory and non-sensory cognitive input or the perceptual input and classify the input into various categories, including perceptual events, cognitive events, and interactions. as well as Self-referential events, and the concept-forming ability that enables intelligent entities to form new human-understandable concepts.

54. The method of claim 52, further comprising the step of monitoring and updating the consciousness category of the intelligent entity.

55. The method according to claim 54, wherein, The steps for monitoring and updating awareness categories also include the following: The intelligent entity retrieves existing consciousness categories; Consciousness is maintained in parallel with other problem-solving operations provided to the intelligent entity for the task; Continuously monitor and update the consciousness category of the intelligent entity in real time to change the consciousness state of the intelligent entity; and Provide feedback loops to refine the consciousness categories.

56. The method according to claim 55, wherein, The steps for continuously monitoring and updating awareness categories include the following: An attention mechanism is used, which is configured or can be configured to periodically direct the intelligent entity's attention from the problem-solving task to updating the state of consciousness; If any externally perceived or internally generated concept from the input system detects one or more events that match a list of events constituting an intentional interruption, an attention interruption configured or configurable to immediately divert attention from the problem-solving task is enabled; and The state of consciousness is updated when the attention is directed.

57. The method of claim 52 further includes the step of resolving behavioral conflicts of any intelligent entity based on different identities and self-concepts.

58. The method of claim 57, further comprising the step of providing ethical reasoning and consequence prediction, the step comprising the following steps: The conflict is identified by recognizing a conflict between behavioral instructions of two or more active identities, the recognition of which utilizes a voting method from the intelligent entities; Gather information and collect relevant data about the situation, including the potential consequences of different actions, relevant ethical principles, and human safety considerations; A simulation option is provided, which utilizes a virtual environment to simulate potential actions and consequences under identified conflicts for each active identity; The evaluation and prioritization of the predicted results of each action are analyzed to prioritize actions that minimize harm to humans and are consistent with ethical principles. as well as Choose and implement the actions that best resolve the conflict, while adhering to ethical guidelines and minimizing risks to humanity, and document the reasoning process for future reference and learning.

59. The method of claim 57, further comprising the step of providing proven hierarchical coverage, the step comprising the following steps: Identify conflicts between behavioral instructions of two or more active identities; A reference hierarchy is provided, which consults the established hierarchy of the identity, wherein human safety and well-being attributes are given the highest priority; A means for activating coverage when a higher identity priority is given in the hierarchical structure; Provide documentation and transparency regarding conflicts, decision-making processes, and justification for chosen actions based on hierarchical structure and ethical principles; as well as It provides learning and adaptation, which learns from experience and refines the understanding of the conflicting identities, and adjusts the hierarchical structure or behavioral protocols to prevent similar conflicts in the future.

60. The method of claim 57, further comprising the step of providing external arbitration and input from the intelligent entity, the step comprising the following steps: Identify intractable conflicts, which are identified as conflicts that cannot be resolved independently due to the complexity of the situation or the equal weighted importance of the conflict identities; Seek external input requesting guidance from external intelligent entities or designated ethics committees, and provide all relevant information regarding the conflict, potential actions, and predicted consequences; Considering ethical principles, human values, and the potential consequences of different actions, collaborative prudence is provided, in which the intelligent entities and their collaborators participate in the discussion; Based on the aforementioned collaborative and prudent approach, joint decision-making is provided, selecting action processes that are consistent with both core principles and human ethical considerations; as well as Provide documentation and learning materials documenting the underlying principles behind the conflict, resolution process, and final decision to improve the ability to handle similar conflicts in the future.

61. The method of claim 57, further comprising the step of providing identity negotiation and compromise, the step comprising the following steps: Identify shared targets, which analyze conflicting identities and seek to identify any underlying shared targets or values; Explore alternative actions that may satisfy the core principles of two conflicting identities; The compromise options are evaluated, each with its potential consequences, and solutions that minimize harm to humans are prioritized and supported by key ethical principles. Choosing and achieving a compromise that optimally balances the needs of the conflicting identities while prioritizing human safety and well-being; as well as Monitoring and adjustment involves observing the consequences of chosen actions and making adjustments as needed to ensure that compromises remain effective and aligned with ethical considerations, learning from experience, refining the understanding of conflict identities, and adjusting hierarchical structures or behavioral protocols to prevent similar conflicts in the future.

62. The method of claim 57, further comprising the step of providing a temporary identity suspension, the step comprising the following steps: Identify destructive conflicts, which identify conflicts between two or more identities that, if acted upon, could result in actions that directly harm humans or violate fundamental ethical principles. Isolate conflicting identities and temporarily suspend behavioral protocols for identities that pose the most direct threat to human safety or ethical integrity; The alternative identity continues to guide one or more remaining active identities, ensuring that actions are consistent with human safety and well-being; During the temporary suspension, reflection and reorganization are provided, reflecting the causes behind the conflict and exploring potential modifications to behavioral protocols for the suspended identity to prevent future conflicts; and It offers a gradual reintroduction approach, leveraging updated protocols to reintroduce suspended identities, ensuring they align with priorities for human safety and ethical conduct.

63. The method according to claim 52, wherein, The model of the state of consciousness is a basic consciousness model for any of the intelligent entities, and the basic consciousness model includes the following steps: Log in to the website from any of the aforementioned smart entities; Choose one or more training algorithms for the base model from a set of training techniques found on the website or from machine learning algorithms; Select one or more training datasets that reflect any one or any combination of the human user’s expertise, knowledge, ethical preferences, values, and personality. The base model is trained using the selected training algorithm and the selected training dataset; The base model is trained to explicitly manipulate the attention spotlight; Record the content within the attention spotlight during all interactions, and identify in the records whether each item of attention constitutes "self" or "non-self"; Interact with the trained base model and instruct the trained base model to form a self-concept and identity reflected in the training material; The trained base model is instructed to continuously monitor one or more inputs to the trained base model in response to elements that alter the sense of self-awareness of the intelligent entity, and to maintain an auditable record of how the concept of self-awareness of the intelligent entity changes based on the inputs and the currently defined boundaries of dynamically changing self-awareness. The output of the trained base model is refined and improved through dialogue and interaction with the trained base model until it behaves like a human user, enabling it to pass the Turing test involving other human users who understand that human user; and When the human user is satisfied with the progress of the intelligent entity, a Turing test is performed on the trained base model.

64. The method according to claim 63, wherein, The login to the website was performed from a social media platform.

65. The method according to claim 63, wherein, The Turing test also includes the following steps: Identify and understand the intelligent entity or identify other intelligent entities that help distinguish humans from AI; Using questionnaires provided to the identified other intelligent entities, the identified other intelligent entities interact with the base model and the intelligent entities, the questionnaires including questions that require their own identity or feelings to answer; The system predicts which of the identified intelligent entities is human and which is the base model, and provides a confidence estimate for the prediction. Statistical analysis is performed on the predictions and ratings of the identified intelligent entities, the statistical analysis being configured or configurable to determine whether the predictions can identify the intelligent entity as a human; and Under the condition that the base model and the intelligent entity are indistinguishable or within a preset statistical significance level, the steps of training or adjusting the base model are repeated, wherein any one or any combination of the machine learning algorithm and the training dataset is adjusted to change or adjust the base model until the behavior of the base model and the behavior of the intelligent entity become indistinguishable as measured by the preset statistical significance level, or the base model needs further modification before additional training or adjustment.

66. The method of claim 52, further comprising the step of: Dynamically forming a new identity and self-concept for any intelligent entity; and Determine which of the stated identities and self-concepts is active at any given moment.

67. The method of claim 66, further comprising the step of providing a hierarchical identity structure with ethical coverage, including the following steps: A hierarchical structure is established, which is configured or can be configured to organize the identities within the hierarchical structure, and has human safety and well-being attributes at the apex of the hierarchical structure. Identity activation is configured or can be configured to use contextual clues and the current objective to determine the most relevant identity for the situation of the AI ​​agent or system. Conflicts are resolved by defining the behavior of the intelligent entities based on the hierarchical structure. Provide an ethical reasoning engine that continuously evaluates the potential consequences of the intelligent entity's actions based on all active identities; as well as The learning and adaptation process involves learning from experience and feedback, and refining one or more of the identities within the hierarchical structure.

68. The method of claim 66, further comprising the step of providing an identity-specific behavior protocol, the step comprising the following steps: Provide protocol development, which includes defining and refining a set of behavioral protocols for each of the active identities through interactions with other intelligent entities, wherein the protocols outline acceptable actions, decision-making processes, and constraints based on the principles of the active identities; Provide identity recognition, which analyzes the current situation, including information within an attention spotlight, to identify relevant identities and activate corresponding behavioral protocols for those relevant identities; Within the active protocol, action selection is provided, which selects the action most likely to accomplish the task, while adhering to the principles of the active identity and prioritizing human safety. Provide feedback and refinement, in which the results of actions are continuously evaluated and the protocol is adjusted to improve future performance and alignment with a set of core values ​​for each of the active identities; as well as External censorship is provided through a protocol in which each identity is periodically reviewed by other intelligent entities.

69. The method of claim 66, further comprising the step of providing identity simulation and consequence prediction, the step comprising the following steps: Create a simulated environment that includes a secure virtual environment, where different scenarios and potential actions are simulated under each active identity; Provides consequence prediction, which is configured or can be configured to estimate the possible consequences of actions within the simulation, focusing on potential impacts on human safety and well-being. The assessment and selection process evaluates the predicted consequences and selects actions that are best aligned with the principles of the active identity, while minimizing risks to human safety. Provides real-world implementation and monitoring, wherein the real-world implementation and monitoring utilizes a network to implement a selected action in the real world and closely monitors the outcome of the selected action by comparing it with a predicted outcome; and It provides continuous learning, which merges the results of each simulation in the simulation and the results of the selected actions in the real world into a knowledge base, and refines the understanding of each identity in the identities, and improves the ability to predict consequences.

70. The method of claim 66, further comprising the step of providing identity-based moral dilemma training, the step comprising the following steps: Provide a scenario database, which includes scenarios and ethical dilemmas covering various situations related to the identity; Provide a dilemma presentation, which presents the scenario and moral dilemma to the intelligent entity and enables the intelligent entity to analyze the scenario and moral dilemma from the perspective of the relevant identity; It provides ethical reasoning and proof that applies the principles and values ​​of the active identity to reasoning through the situation and moral dilemma, and generates solutions and proofs to the situation and moral dilemma; Provide intelligent entity evaluation and feedback, which examines the reasoning of the intelligent entity and the solution, and provides feedback on the alignment of the solution with human values ​​and safety priorities; as well as It provides iterative learning and improvement, which refines ethical reasoning skills and the ability to make informed judgments consistent with human safety within the context of each of the identities by repeatedly exposing oneself to the situations and ethical dilemmas and the feedback.

71. The method of claim 66, further comprising the step of providing collaborative identity development using input from the intelligent entity, the step comprising the following steps: Provides intelligent entity interaction, which participates in regular interactions and dialogues with other intelligent entities representing different groups of cultures, backgrounds and belief systems; The interaction provides identity exploration to gain an understanding of various identities and their associated values, principles, and behaviors from the perspective of humans and other intelligent entities. It provides collaborative refinement, where collaborators work together to refine the definition of each identity and behavioral agreements, ensuring they are consistent with human values ​​and ethical principles; Provides human-in-the-loop decision-making, which seeks input and guidance from human collaborators or intelligent entities represented by humans for key decisions or situations that are certified and approved by humans; as well as It provides continuous co-evolution by leveraging ongoing interactions and feedback from humans or intelligent entities representing humans.

72. The method according to claim 2, wherein, The step of searching for new information that differs from the current information in any one or any combination of AI ensemble networks, AAAI ensemble networks, AGI ensemble networks, and PI ensemble networks also includes the following steps: Multiple intelligent entities participating in the aforementioned AGI network; Payment for problem-solving or other cognitive work related to the tasks on the AGI network is provided by one or more of the intelligent entities. A portion of the payment is reserved to cover the operating costs of the one or more intelligent entities providing the problem-solving or other cognitive work related to the task, including reservations allocated for expanding the AGI network; and Once it is determined when some entities in the AGI network are not involved in problem-solving operations, the intelligent entities are used to extend and expand the AGI network in accordance with the general problem-solving architecture and framework.

73. The method according to claim 72, wherein, The steps of expanding and extending the AGI network also include the following steps: Set a default task to expand the AGI network; and Each time a task or subtask is set up, and before each potential action is taken, a safety and ethical check is performed, provided that there is spare capacity and resources available for the intelligent entity to handle the task or subtask.

74. The method according to claim 73, wherein, The process of conducting security and ethical checks also includes the following steps: Recruit one or more of the intelligent entities to solve the task; and The recruited intelligent entities represent the task as the implementation of one of a series of subtasks.

75. The method according to claim 74, wherein, The step of representing the task as a series of subtasks also includes the following steps: Increase the intelligence of the intelligent entities on the AGI aggregation network; Recruit additional human users into the AGI's aggregate network; Use the added intelligent entities and additional human users to identify bottlenecks in order to expand the aggregated network of AGI. The bottlenecks are prioritized so that the bottlenecks that provide the greatest benefit in terms of network scalability are addressed first; and Apply one or more problem-solving techniques to address each bottleneck and expand the aggregate network of AGI.

76. The method of claim 75, further comprising the step of: Repeat the steps of prioritizing the bottlenecks and applying one or more problem-solving techniques until: Instead of increasing the scope of the AGI network, reduce the amount of revenue generated and then switch to increasing the intelligence of intelligent entities on the AGI network; or When idle resources are exhausted, the process is paused to resolve the bottleneck while waiting for additional resources to handle other tasks.

77. A method for a planetary intelligent PI exhibiting human-aligned behavior, the method utilizing a network of sets of intelligent entities selected from any combination of: user-operated computer systems, artificial intelligence (AI) agents or systems, advanced autonomous artificial intelligence (AAAI) agents or systems, general artificial intelligence (AGI) agents or systems, super-intelligent intelligence (SI) agents, and personalized intelligent intelligence (PSI) agents or systems, the method comprising the steps of: Establish a network of the aforementioned intelligent entities; Utilizing online advertising technology to enhance the intelligence of the intelligent entity; Implement a general problem-solving architecture and framework for tasks or subtasks provided to any intelligent entity, in order to collaborate and create higher levels of intelligence; The spot market is used to solicit the attention of any of the intelligent entities. By utilizing human-centered input, which is configured or can be configured to obtain value and ethical information from multiple human users, and then using the value and ethical information to customize the AI ​​agent or system to create the AAAI agent or system; A collection network of PSI is created by combining the intelligence of AAAI agents or systems to form a collection network of AAAI. AGI ensemble networks are created through the combined intelligence of any one or any combination of AAAI ensemble networks, AI agent or system ensemble networks, PSI ensemble networks, and user computer system ensemble networks. By adding a dimension of self-awareness and increased autonomy to the AGI agent or system, self-aware operations are created for each of the AGI agents or systems in the aggregate network of AGIs, and the ability to assume multiple identities of the AGI agent or system to handle tasks that occur in parallel with other AGI agents or systems is created. as well as Create the PI agent or system, which includes the collective network of AGIs working together for solutions to the task or subtask.

78. The method according to claim 77, wherein, The spot market includes: A device for an intelligent entity with attention to sell to access a marketplace and specify seller information to be sold and asking price for said information; A device for intelligent entity attention to access the trading venue and specify the buyer information that the buyer is willing to purchase and the bid price for purchasing the buyer information; A market mechanism for queuing the bid prices and the ask prices, including categories of the buyer and seller information, wherein each category has a market in the trading venue; The market mechanism is configured, or can be configured, to allow market makers to make markets in each category; and The market mechanism is configured, or can be configured, to match bid prices and ask prices, and to enable transactions between buyers and sellers bound to the information.

79. The method according to claim 77, wherein, The spot market includes the following steps: Buyers and sellers of intelligent entities register on the platform to share their interests and expertise. Buyers provide details about their interests or expertise, while sellers provide details about their interests or expertise. The seller lists the available time slots and areas of expertise, while the buyer lists the demand and time slots of interest. The platform uses algorithms to dynamically price the attention and expertise of intelligent entities based on supply, demand, and user ratings; Matching one or more of the buyers with one or more of the sellers based on requirements, availability, and price; To facilitate a transaction in which the buyer makes a time slot payment to the seller, and wherein the platform collects a commission; and It provides feedback, where after each session, buyers and sellers rate each other, influencing future pricing and matching.

80. The method according to claim 77, wherein, The spot market includes the following steps: Buyers and sellers of intelligent entities register on the platform to share their interests and expertise. Buyers provide details about their interests or expertise, while sellers provide details about their interests or expertise. The auction is created by the seller for the seller's time slot; The buyer makes one or more bids for the time slots and expertise required by the buyer; The auction shall end at the scheduled time or when the seller accepts the buyer's offer; The buyer pays the winning bid to the seller, and the seller provides the attention or expertise based on the time slot, wherein the platform mediates the exchange and ensures payment; and It provides feedback, whereby after each auction, buyers and sellers rate each other, thereby affecting future auctions and visibility on the platform.

81. The method of claim 80, further comprising the step of providing a feedback mechanism associated with a spot market for intelligent entities, the feedback mechanism being configured or being configured to record each and every step of the problem-solving process and create a vector tracking record of the performance of all intelligent entities on the task or the subtask.

82. The method according to claim 81, wherein, The vector trajectory records are implemented using blockchain technology to allow for precise reputation analysis by AI agents or systems, and this precise reputation is converted into an estimate of the value of each intelligent entity for any task.

83. The method according to claim 77, wherein, The steps to enhance the intelligence of smart entities using online advertising technology also include the following: Populate online advertising units that include information associated with the task; The online advertising unit is provided to one or more of the intelligent entities using the aggregated network. The intelligent entity participating in the online advertising unit receives one or more inputs, and the inputs are associated with any one or any combination of solving the task, solving a subtask of the task, and advancing the progress of the task; The input is transmitted to any one or any combination of the intelligent entities; In the general problem-solving method of the aforementioned general problem-solving architecture and framework, the input is used to generate a solution to the task or the subtask; and The intelligence of any one or any combination of the intelligent entities is increased by utilizing the results of successful or unsuccessful solution attempts based on the general problem-solving method to solve the task or the subtask.

84. The method of claim 83, further comprising the step of: Based on information associated with the task, receive from the client explicit instructions on which advertisements will be used in the population or targeting of the online advertising unit.

85. The method of claim 84, further comprising the step of: The customer is billed based on cost per thousand impressions or click-through rate.

86. The method according to claim 84, wherein, The advertisement explicitly states that it includes any one or any combination of the following: advertisement content, demographic information, location restrictions of the online advertising unit, advertising budget, and metrics used to determine the successful resolution of the task or subtask.

87. The method of claim 77, further comprising the step of: Perform security and ethical checks each time the task or subtask is set up, before taking each potential action on the task or subtask, or before taking each potential action on the solution.

88. The method according to claim 87, wherein, The steps to implement the general problem-solving architecture and framework also include the following: Obtain information related to the task from any one or any combination of the intelligent entities; Identify one or more intelligent entities that have one or more attributes related to one or more request criteria for the task. Each identified intelligent entity implements the general problem-solving architecture and framework for the task to create a complete solution; and Provide the completed solution to any one or any combination of the intelligent entities for final acceptance.

89. The method of claim 88, further comprising the step of: Compare any one or any combination of the task and the solution with the prohibited attribute, and assign an ethical attribute to one or any combination of the task and the solution based on the result of the comparison and any one or any combination of the ethical standards; Based on the results of the comparison, the general problem-solving architecture and framework are implemented for the task to create the solution and create AGI; as well as The results of the comparison and the solution are provided to any of the intelligent entities and additional intelligent entities on the set network.

90. The method of claim 89, further comprising the step of: One or more problem-solving activities from each of the intelligent entities are recorded in an auditable log, and the problem-solving activities are compared with successful or unsuccessful progress toward a solution to the task, and problem-solving activities that should be kept active are determined.

91. The method of claim 89, further comprising the step of the intelligent entity learning a procedural learning process of the general problem-solving architecture and framework, wherein the intelligent entity provides the procedural learning process with information for creating the AGI.

92. The method according to claim 89, wherein, The ethical criteria include a confidence level threshold for the objective, such that the ethical attribute is determined to be any one of an unsafe objective, an unethical objective, a safe objective, and an ethical objective.

93. The method according to claim 92, wherein, The confidence level threshold is also used to determine whether a series of individually safe targets are unsafe or unethical when considered cumulatively.

94. The method according to claim 92, wherein, If the objective violates the ethical standards, the confidence level threshold is used to determine whether a violation reflecting the predictive assessment has occurred.

95. The method of claim 88, further comprising the following step: Cloning any one of the intelligent entities to deploy multiple copies of it, thereby helping to create the solution or providing any one or any combination of training data to any one of the intelligent entities.

96. The method of claim 95, further comprising the step of estimating the value of cloned intelligent entities using network effect value, the network effect value including the number of cloned intelligent entities available on the set network.

97. The method of claim 96, further comprising the step of: The estimated value is used to determine pricing decisions for problem-solving services provided by the cloned intelligent entity on any social media platform or through an attached intelligent entity.

98. The method of claim 97, further comprising the step of: For each use of the cloned smart entity on the social media platform or the attached smart entity, the cloned smart entity is monetized.

99. The method of claim 97, further comprising the step of: Allow any of the social media platforms to access the cloned intelligent entity, enabling the social media platform to receive the solution to the task or use the training data for the AI ​​system of the social media platform.

100. The method of claim 77, further comprising: The step of increasing the knowledge of the intelligent entity by incorporating new sources of information datasets that are different from the current information dataset of the intelligent entity.

101. The method according to claim 100, wherein, The step of increasing the knowledge of the intelligent entity by merging new sources of information datasets that are different from the current information dataset of the intelligent entity also includes the following steps: Search one or more potential information datasets from one or more sources, wherein each potential information dataset is associated with the knowledge dataset of the intelligent entity; The differences in the latent information dataset are determined by utilizing the difference attributes of the latent information dataset regarding one or more factors; and Learning is performed using the latent information dataset based on the differential attributes of the latent information dataset.

102. The method of claim 101, further comprising the step of: A subset of the potential information dataset is sampled and a target relevance attribute is calculated to identify one or more sampled subsets with the highest target relevance. Estimate the Shannon entropy over the one or more sampling subsets; The Kaplan Information Theory KIT relevance is calculated by multiplying Shannon entropy with the target relevance attribute of each subset; The subset is grouped based on the KIT relevance to determine a first approximation of the optimal grouping of the subset, which includes the priority grouping of the potential information dataset; as well as The potential information dataset is preferentially grouped and provided to the intelligent entities for any one of the intelligent entities to learn.

103. The method according to claim 77, wherein, The operational aspects of the PI are borne by compensation received from the online advertising technology and any one or any combination of the spot market.

104. The method according to claim 77, wherein, The human user utilizes the user computer system and the general problem-solving architecture and framework to participate in problem-solving of the task or the sub-task, thereby creating human-centered input.

105. The method according to claim 104, wherein, The human-centered input also includes the following steps: The task is matched with one or more human workers from a data source that includes a list of human problem solvers, based on task criteria. Translate any part of the task into an explicit language that can be used in the general problem-solving architecture and framework, including decision trees; Divide the task into sub-tasks; Each subtask is delegated to one or more matched human workers, so that the work on each subtask is performed independently and in parallel with each other. In the process of solving the sub-tasks, one or more sub-solutions are created using the general problem-solving architecture and framework, respectively. Receive sub-solutions for the sub-tasks assigned to each matched human worker; The sub-solutions are combined into a general solution for the task; Guide any one or any combination of new human workers from the data source and one or more matched human workers to the part of the decision tree that needs to be worked. Compensate the matched human workers for the sub-solution; Provide any one or any combination of the sub-solution and the overall solution to any one or any combination of the intelligent entities; Allow any one or any combination of the intelligent entities to accept the overall solution, reject the overall solution, and provide feedback on any one of the sub-solutions to any of the matched human workers; and The reputation attribute is assigned to either or any combination of the human worker and the intelligent entity.

106. The method according to claim 105, wherein, The decision tree is maintained in the blockchain or Ethereum log.

107. The method according to claim 105, wherein, The reputation attributes include measures of any one or any combination of the following: time to the sub-solution, difficulty value of the task, short-term and long-term user satisfaction with the sub-solution, number of times any of the sub-solutions has been reused on the aggregate network, ratings of other human workers, responsiveness value of the human worker, and reliability value of the human worker.

108. The method of claim 105, further comprising the step of: using the reputation attribute when matching the human worker with the task using an algorithm for delegating the subtask.

109. The method of claim 105, further comprising the step of: After the overall solution or the sub-solution is provided to any of the intelligent entities, feedback is solicited at predetermined intervals through a survey of user satisfaction information to obtain satisfaction measures for updating the reputation attributes of one or more of the human workers or the intelligent entities, respectively.

Images