Information processing method based on artificial intelligence, device, electronic equipment, and agent

Abstract

To provide an information processing method capable of easily and efficiently improving the degree of agent intelligentization, a device, electronic equipment, an agent, a storage medium, and a program.SOLUTION: An information processing method based on AI includes acquiring input information to be processed, and determining execution information related to input information processing. The execution information includes at least one of memory information to be retrieved or tool information to be invoked. The method further includes acquiring, by using the execution information, at least one processing result information corresponding to input information processing, and generating output information for feedback by integrating at least processing result information.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present disclosure relates to the field of artificial intelligence (AI), in particular to the field of large-scale language models (LLMs), AI agents, etc., and specifically to an information processing method, device, electronic device, computer-readable storage medium, computer program product, and AI agent based on artificial intelligence. [Background technology]

[0002] Artificial intelligence is a field that studies how computers can imitate some of the human thought processes and intellectual behaviors (e.g., learning, reasoning, thinking, planning, etc.), and includes both hardware and software technologies. AI hardware technologies generally include sensors, AI-dedicated chips, cloud computing, distributed storage, big data processing, etc., while AI software technologies mainly include computer vision technology, voice recognition technology, natural language processing technology, machine learning / deep learning, big data processing technology, knowledge graph technology, and other major directions.

[0003] Recently, with the continuous progress of large-scale language models, the capabilities of AI in language understanding, knowledge memory, logical reasoning and other areas have been greatly improved, which strongly supports the development of AI agents. AI agents are advanced artificial intelligence systems with large-scale language models as their core inference engine. They not only have the language understanding and generation capabilities of large-scale language models, but can also solve various complex problems efficiently and flexibly, and further release the machine intelligence contained in large-scale language models to provide users with more accurate and personalized services.

[0004] The methods described in this section are not necessarily methods that have been previously conceived or adopted. Unless otherwise noted, any methods described in this section should not be considered prior art merely by virtue of their inclusion in this section. Similarly, unless otherwise noted, any problem addressed in this section should not be considered to be an admission of prior art. Summary of the Invention

[0005] The present disclosure provides an artificial intelligence-based information processing method, apparatus, electronic device, computer-readable storage medium, and computer program product, as well as an AI agent.

[0006] According to one aspect of the present disclosure, an artificial intelligence based information processing method is provided, including: obtaining input information to be processed; determining execution information related to the processing of the input information, the execution information including at least one of memory information to be searched or tool information to be recalled; using the execution information to obtain at least one processing result information corresponding to the processing of the input information; and integrating the at least one processing result information to generate output information for feedback.

[0007] According to another aspect of the present disclosure, there is provided an AI agent configured to perform the above method.

[0008] According to another aspect of the present disclosure, an artificial intelligence based information processing apparatus is provided, comprising: an input unit configured to obtain input information to be processed; a planning unit configured to determine execution information related to processing of the input information, the execution information including at least one of memory information to be searched or tool information to be recalled; an action unit configured to utilize the execution information to obtain at least one processing result information corresponding to the processing of the input information; and an output unit configured to integrate the at least one processing result information to generate output information for feedback.

[0009] According to another aspect of the present disclosure, an electronic device is provided, the electronic device including at least one processor and a memory communicatively coupled to the at least one processor, the memory storing instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method described above.

[0010] According to another aspect of the present disclosure, a non-transitory computer readable storage medium having stored thereon computer instructions is provided, where the computer instructions are used to cause a computer to perform the above-described method.

[0011] According to another aspect of the present disclosure there is provided a computer program product including a computer program which, when executed by a processor, implements the method described above.

[0012] According to one or more embodiments of the present disclosure, the intelligence level of an AI agent can be increased simply and efficiently.

[0013] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, and are not intended to limit the scope of protection of the present disclosure. Other features of the present disclosure will be easily understood from the following description. [Brief description of the drawings]

[0014] The drawings illustratively illustrate the examples, constitute a part of the specification, and together with the written description serve to explain exemplary embodiments of the examples. The illustrated examples are for illustrative purposes only and are not intended to limit the scope of the claims. In all the drawings, the same reference numerals refer to similar, but not necessarily identical, elements. [Figure 1] FIG. 1 is a schematic diagram illustrating an example system capable of implementing the methods described herein, according to an embodiment of the present disclosure. [Diagram 2]FIG. 2 illustrates a flowchart of an artificial intelligence-based information processing method according to one embodiment of the present disclosure. [Diagram 3] FIG. 2 illustrates a flowchart of an artificial intelligence-based information processing method according to another embodiment of the present disclosure. [Figure 4] 4 is a flow chart illustrating a process for determining performance information associated with processing input information according to an embodiment of the present invention. [Diagram 5] FIG. 2 is a schematic diagram of an AI agent according to an embodiment of the present disclosure. [Figure 6] FIG. 1 is a block diagram showing a configuration of an information processing device based on artificial intelligence according to an embodiment of the present disclosure. [Figure 7] FIG. 13 is a block diagram showing a configuration of an information processing device based on artificial intelligence according to another embodiment of the present disclosure. [Figure 8] FIG. 1 is a block diagram showing a configuration of an exemplary electronic device for implementing an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] Hereinafter, exemplary embodiments of the present disclosure will be described in conjunction with the drawings, and various details in the embodiments of the present disclosure contained therein are to be considered as merely illustrative, in order to facilitate understanding. Therefore, it should be recognized that those skilled in the art can make various changes and modifications to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, the following description omits descriptions of known functions and structures.

[0016] In this disclosure, unless otherwise specified, the terms "first," "second," and the like, used to describe various elements are not intended to limit the location, timing, or importance of these elements. Such terms are used only to distinguish one element from another. In some examples, a first element and a second element may refer to the same instance of the element, or in some cases, may refer to different instances based on the contextual description.

[0017] The terms used in the description of the various above-mentioned examples of the present disclosure are intended only to describe specific examples and are not intended to be limiting. Unless otherwise clearly indicated by the context, the elements may be one or more unless specifically limited in number. It should be noted that the term "and / or" used in the present disclosure covers any and all possible combinations of the listed items.

[0018] Among related technologies, existing AI agents are still in a fairly early stage of development, and there is no unified standard for the design of agent mechanisms related to AI agents. Currently, many of the various AI agents appearing in the industry are still in the proof-of-concept and early trial stages.

[0019] For example, existing AI agents generally suffer from the following problems:

[0020] First, the overall architecture of the AI ​​agent has the problem of being too coarse-grained or too fine-grained in its module design. If the module is too coarse-grained, it may place excessive cognitive load on one module, thereby limiting the gains that the AI ​​agent can bring compared to large-scale language models. If the module is too fine-grained, the task-solving link may be long and the accumulated error may be large, thus failing to effectively utilize the language understanding capabilities of the large-scale language model.

[0021] And AI agents usually incorporate too many artificially set rules to adapt to specific problems in specific scenes. However, when they encounter a new problem, the design that relies on these preset rules may limit the AI ​​agent's ability to respond in unknown environments and prevent it from flexibly adjusting its strategy, resulting in poor generalization performance in real applications.

[0022] In addition, the realization of specific functions of AI agents is still insufficient. Some AI agents adopt vector databases to simulate the memory function of the human brain and store all historical interaction behaviors in the vector database. Although this seems to achieve lossless long-term memory, it prevents the AI ​​agent from extracting reusable abstracted experiences from a series of conversations, tasks, and events. The AI ​​agent can only record, but cannot summarize, which leads to a performance bottleneck on memory.

[0023] Moreover, AI agents do not have more advanced cognitive capabilities, such as the ability to second guess, and the results currently obtained may not be an upper limit of the capabilities of the AI ​​agents themselves, even if they meet the basic requirements of the task.

[0024] In response to at least one of the problems described above, embodiments of the present disclosure provide an information processing method and an AI agent based on artificial intelligence.

[0025] Before describing the methodologies of the embodiments of the present disclosure in detail, an exemplary system in which the methods described herein may be implemented will first be described with reference to FIG.

[0026] Figure 1 is a schematic diagram of an example system 100 in which various methods and apparatus described herein may be implemented, according to embodiments of the present application. With reference to Figure 1, the system 100 includes one or more client devices 101, 102, 103, 104, 105, and 106, a server 120, and one or more communication networks 110 coupling the one or more client devices to the server 120. The client devices 101, 102, 103, 104, 105, and 106 may be configured to execute one or more applications.

[0027] In an embodiment of the present disclosure, server 120 may execute one or more services or software applications that enable the execution of artificial intelligence-based information processing methods or AI agents.

[0028] In some embodiments, server 120 may also provide other services or software applications, which may include non-virtualized and virtualized environments. In some embodiments, these services may be provided as web-based or cloud services, for example, provided to users of client devices 101, 102, 103, 104, 105, and / or 106 in a Software-as-a-Service (SaaS) model.

[0029] In the arrangement shown in FIG. 1, server 120 may include one or more assemblies that implement the functions performed by server 120. These assemblies may include software assemblies, hardware assemblies, or a combination thereof that may be executed on one or more processors. Users operating client devices 101, 102, 103, 104, 105, and / or 106 may in turn use one or more client applications to interact with server 120 to utilize the services provided by these assemblies. It should be understood that a variety of different system arrangements are possible and may differ from system 100. Thus, FIG. 1 is an example of a system for implementing various methods described herein and is not intended to be limiting.

[0030] A user may use client devices 101, 102, 103, 104, 105, and / or 106 to provide information for interacting with an AI agent. The client devices may provide an interface through which a user of the client device interacts with the client device. The client devices may also output information to the user through the interface. Although only six client devices are illustrated in FIG. 1, one skilled in the art will appreciate that the present disclosure may support any number of client devices.

[0031] The client devices 101, 102, 103, 104, 105, and / or 106 may include various types of computing devices, such as portable handheld devices, general purpose computers (e.g., personal computers or laptops), workstation computers, wearable devices, smart screen devices, self-service terminal devices, service robots, gaming systems, thin clients, various messaging devices, sensors, or other sensing devices. These computing devices may run various types and versions of software applications and operating systems, such as MICROSOFT Windows, APPLE iOS, UNIX-like operating systems, Linux or Linux-like operating systems (e.g., GOOGLE Chrome OS), and may include various mobile operating systems, such as MICROSOFT Windows Mobile OS, iOS, Windows Phone, Android, and the like. Portable handheld devices may include mobile phones, intelligent phones, tablets, personal digital assistants (PDAs), and the like. Wearable devices may include head-mounted displays (e.g., smart glasses) and other devices. Gaming systems may include various handheld gaming devices, Internet-enabled gaming devices, and the like. The client devices may run a variety of applications, such as Internet-related applications, communication applications (eg, email applications), Short Message Service (SMS) applications, and may use a variety of communication protocols.

[0032] Network 110 may be any type of network known to those skilled in the art that may use any one of a number of available protocols (including, but not limited to, TCP / IP, SNA, IPX, etc.) to support data communications. By way of example, one or more networks 110 may be a local area network (LAN), an Ethernet-based network, a token loop, a wide area network (WAN), the Internet, a virtual network, a virtual private network (VPN), an intranet, an extranet, a blockchain network, a public switched telephone network (PSTN), an infrared network, a wireless network (e.g., Bluetooth, WIFI), and / or any combination of these and / or other networks.

[0033] Server 120 may include one or more general purpose computers, dedicated server computers (e.g., PC (personal computer) servers, UNIX servers, mid-range servers), blade servers, mainframe computers, server clusters, or any other suitable arrangement and / or combination. Server 120 may include one or more virtual machines running a virtual operating system, or other computing architecture involving virtualization (e.g., one or more flexible pools of virtualized logical storage devices to maintain the server's virtual storage devices). In various embodiments, server 120 may execute one or more services or software applications that provide the functionality described below.

[0034] The computing units in server 120 may run one or more operating systems, including any of the operating systems listed above and any commercially available server operating system. Server 120 may also run any one of a variety of additional server applications and / or mid-tier applications, such as an HTTP server, an FTP server, a CGI server, a JAVA server, a database server, etc.

[0035] In some embodiments, server 120 may include one or more applications for analyzing and consolidating data feeds and / or event updates received from users of client devices 101, 102, 103, 104, 105, and / or 106. Server 120 may include one or more applications for displaying data feeds and / or real-time events via one or more display devices of client devices 101, 102, 103, 104, 105, and / or 106.

[0036] In some embodiments, the server 120 may be a server of a distributed system or a server incorporating blockchain. The server 120 may be a cloud server, or an intelligent cloud computing server or an intelligent cloud host equipped with artificial intelligence technology. The cloud server is a host product in a cloud computing service system, and solves the defects of the traditional physical host and virtual private server (VPS) services, such as high management difficulty and low business scalability.

[0037] System 100 may also include one or more databases 130. In some embodiments, these databases may be used to store data or other information. For example, one or more of databases 130 may be used to store information such as audio files or video files. Databases 130 may be located in a variety of locations. For example, a database used by server 120 may be local to server 120 or may be remote from server 120 and in communication with server 120 over a network or dedicated connection. Databases 130 may be of a variety of types. In some embodiments, a database used by server 120 may be a relational database. One or more of these databases may store, update, and retrieve data from the database and from the database in response to instructions.

[0038] In some embodiments, one or more of databases 130 may be used by an application to store data for the application. The databases used by the application may be various types of databases, such as a key-value repository, an object repository, a generic repository supported by a file system, etc.

[0039] The system 100 of FIG. 1 can be configured and operated in a variety of ways to enable application of the various methods described according to the present disclosure.

[0040] Various aspects of an artificial intelligence-based information processing method and an AI agent according to embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings.

[0041] FIG. 2 illustrates a flow chart of an artificial intelligence based information processing method 200 according to one embodiment of the present disclosure.

[0042] In an embodiment of the present disclosure, the artificial intelligence method 200 may be executed by an AI agent. For example, the AI ​​agent may execute the artificial intelligence-based information processing method 200 during interaction with a user or an external environment.

[0043] As shown in FIG. 2, the method 200 includes steps S202, S204, S206, and S208.

[0044] In step S202, input information to be processed is obtained.

[0045] In one example, the input information may include inquiry information, request information, command information, signal information, data information, etc. The input information may be provided in the process of a user interacting with an AI agent, or may be obtained by the AI ​​agent sensing the external environment. In application scenarios such as a smart album butler, a smart voice assistant, and a smart cockpit, the input information may include photos, videos, voice, text, vehicle sensor data, etc. The input information may also include context information such as time, location, object, and situation.

[0046] In step S204, execution information associated with processing the input information is determined, the execution information including at least one of stored information to be retrieved or tool information to be invoked.

[0047] In one example, the execution information may be information that the AI ​​agent needs to additionally use, utilize, or assist in processing the input information. In one case, the AI ​​agent may determine that it only needs to look up memory information so that it can process the input information. In another case, the AI ​​agent may determine that it only needs to call up tool information so that it can process the input information. In yet another case, the AI ​​agent may determine that it needs both to look up memory information and to call up tool information so that it can process the input information.

[0048] That is, an AI agent according to an embodiment of the present disclosure has planning capabilities to determine what appropriate execution information to use to process input information.

[0049] In one example, memory information may refer to memory contents that an AI agent has. Memory information may be stored in the form of a vector database. Memory information may also be maintained for personalized content, such as a user image, user preferences, etc., for a particular user.

[0050] That is, an AI agent according to an embodiment of the present disclosure also has memory storage capabilities.

[0051] In one example, tool information can refer to the names or types of tools that the AI ​​agent can use, connect to, and access, such as a calculator for performing mathematical operations, a search engine for performing searches, the python language for performing data analysis, etc.

[0052] That is, an AI agent according to an embodiment of the present disclosure also has computing capabilities.

[0053] In step S206, the execution information is used to obtain at least one piece of processing result information corresponding to the processing of the input information.

[0054] In one example, when the AI ​​agent determines that it needs to search for memory information, a query vector can be generated as an input to a memory retriever, so that one or more relevant memory information retrieved from all memory information held by the AI ​​agent is output through the memory retriever. In this case, using the memory information to obtain at least one processing result information corresponding to processing the input information can include searching the memory information related to the input information to obtain at least one memory information search result.

[0055] In one example, when an AI agent determines that it needs to retrieve tool information, it can first quickly find potentially useful candidate tool information through a lightweight vector retriever, and then determine one or more related tool information from the candidate tool information, thereby quickly finding appropriate tool information among a large amount of tool information.

[0056] In one example, when calling up the tool information to process the input information, corresponding usage parameters are filled in according to the usage instructions of the tool information, so as to realize the processing of the input information using the tool information. In this case, using the tool information to obtain at least one processing result information corresponding to the processing of the input information may include using the tool information to complete the processing of the input information and obtain at least one tool information calling result, such as using a calculator to calculate a formula provided by a user, using Python to perform data analysis on data provided by a user, and using a search engine to make a weather forecast for the weather requested by a user. That is, the AI ​​agent according to the embodiment of the present disclosure has the ability to act. Once planning is completed, the AI ​​agent can begin to act. In the embodiment of the present disclosure, the planning ability and the ability to act of the AI ​​agent can be similar to the concept of planning and execution included in the PDCA (Plan, Do, Check, Act) loop in business administration.

[0057] In step S208, the at least one processing result information is integrated to generate output information for feedback.

[0058] In one example, the integrating operation may refer to further processing or processing the at least one processing result information to generate feedback to a user (e.g., a response to a user) or feedback to an external environment (e.g., an action or response to an external environment). For example, integrating the at least one processing result information may include selecting one or more results from the at least one stored information search result and / or the at least one tool information call result, or aligning the at least one stored information search result and / or the at least one tool information call result to a specific result. In the embodiments of the present disclosure, the integrating operation may vary according to actual situations or may be determined by the inference ability underlying the large-scale language model, and the present disclosure is not limited thereto.

[0059] In one example, the output information can be styled according to the scene in which the method 200 is applied. For example, a voice scene such as a smart home can announce to the user in a helpful, friendly, colloquial language. As another example, the output information can be multimodal, for example interweaving the output text with images to form a pictorial and textual effect.

[0060] The information processing method 200 based on artificial intelligence according to the embodiment of the present disclosure provides a simple and effective AI agent mechanism for improving the intelligence level of an AI agent. In this AI agent mechanism, the AI ​​agent is provided with a search function for memory and a call function for tools. In addition, the AI ​​agent, as a single initiator, starts to use memory information and / or tool information through the planning ability in step S204 and the behavioral ability in step S206, and receives the corresponding usage results, so that the control ability including the planning ability and the behavioral ability of the AI ​​agent only focuses on the main control function aspect, and does not redundantly mix in the additional function aspects of memory capacity and / or computing capacity, and at the same time, there is no communication coupling between memory capacity and computing capacity, thereby realizing a highly cohesive modular function design and a low-coupling communication mechanism design, thereby simply and effectively improving the intelligence level of the AI ​​agent.

[0061] In addition, since the artificial intelligence-based information processing method 200 according to the embodiment of the present disclosure is not designed depending on a specific or particular application scenario, a general-purpose AI agent mechanism that can be used in a variety of application scenarios can be realized, thereby improving the flexibility and versatility of the AI ​​agent and enhancing the degree of customization of the AI ​​agent.

[0062] In the technical solution disclosed herein, the collection, storage, use, processing, transmission, provision and disclosure of relevant user personal information shall all comply with the provisions of relevant laws and regulations and shall not violate public order and morals.

[0063] Various aspects of the artificial intelligence based information processing method according to embodiments of the present disclosure are described in further detail below.

[0064] FIG. 3 shows a flow chart of an artificial intelligence based information processing method 300 according to another embodiment of the present disclosure.

[0065] As shown in Fig. 3, the method 300 may include steps S302, S304, S306, S307, and S308. Note that the operations of steps S302, S304, S306, and S308 may be the same as steps S202, S204, S206, and S208 shown in Fig. 2, and therefore detailed descriptions thereof will be omitted.

[0066] In step S302, input information to be processed may be obtained.

[0067] In step S304, execution information associated with processing the input information may be determined, the execution information including at least one of stored information to be retrieved or tool information to be invoked.

[0068] In step S306, the execution information can be utilized to obtain at least one piece of processing result information corresponding to the processing of the input information.

[0069] In step S307, it can be determined whether at least one processing result information meets a preset criterion. In one example, step S307 can be performed based on the inference ability underlying a large-scale language model.

[0070] Meanwhile, in response to determining that the at least one processing result information meets the preset criterion ("Yes" shown in FIG. 3), the method 300 may proceed to step S308.

[0071] In step S308, the at least one processing result information is integrated to generate output information for feedback.

[0072] On the other hand, in response to determining that the at least one processing result information does not meet the preset criteria ("NO" shown in FIG. 3), the method 300 returns from step S307 to step S306, and reuses the execution information to obtain at least one processing result information corresponding to the processing of the input information.

[0073] Then, in step S307, it is again determined whether the at least one processing result information meets the preset criterion, so that the method 300 either proceeds to step S308 or returns to step S306 to repeat the above-mentioned process.

[0074] In one example, when the method 300 is applied to a search expansion scene, the AI ​​agent may dynamically change search terms according to search results during the search process to obtain results that best match the user's intention.

[0075] In this manner, the AI ​​agent according to the embodiment of the present disclosure may have an evaluation capability. When the AI ​​agent obtains at least one memory information search result and / or at least one tool information call result, the AI ​​agent may determine whether the result is as expected or not. The evaluation capability may prevent the AI ​​agent from accumulating excessive errors when dealing with complex problems or from entering a dead loop in the task process.

[0076] Meanwhile, the AI ​​agent according to the embodiment of the present disclosure may also have a rethinking capability: if the AI ​​agent determines that at least one stored information search result and / or at least one tool information call result is unexpected, it will modify the previous process of utilizing the execution information (e.g., the usage parameters previously inputted when calling the tool information) to reacquire the result.

[0077] As mentioned above, in the embodiments of the present disclosure, the planning, acting, evaluating, and rethinking capabilities of an AI agent can be similar to the concept of plan, do, check, and act included in the PDCA loop in business management.

[0078] Therefore, the artificial intelligence-based information processing method 300 according to the embodiment of the present disclosure further incorporates evaluation ability and reconsideration ability into the control ability of the AI ​​agent, so that the AI ​​agent improves its upper limit of ability through self-evaluation and self-reconsideration, thereby further improving the intelligence level of the AI ​​agent.

[0079] In some embodiments, since the input information may be multimodal, step S202 shown in FIG. 2 or step S302 shown in FIG. 3 may include converting the multimodal input information into a unified information format.

[0080] In this way, the input information can be distributed losslessly for subsequent processing, thereby ensuring correct processing actions by the AI ​​agent.

[0081] FIG. 4 illustrates a flow diagram of a process 400 for determining performance information associated with processing of input information, according to an embodiment of the present disclosure.

[0082] In one example, the process 400 may correspond to step S204 shown in FIG. 2 or step S304 shown in FIG.

[0083] As shown in FIG. 4, the process 400 may include steps S402, S404, and S406 or S408.

[0084] In step S402, at least one task associated with processing the input information may be determined based on the input information.

[0085] For each task of the at least one task, in step S404, it can be determined whether the task needs to be processed based on the execution information.

[0086] In one example, the AI ​​agent may first consider whether it needs to retrieve memory information and / or call upon tool information to handle the task in question, a process that may be performed based on the inference capabilities underlying large-scale language models.

[0087] In some embodiments, in response to determining that the corresponding task needs to be processed based on the execution information, execution information corresponding to the corresponding task can be determined in step S406. This process can be performed based on the inference capabilities underlying the large-scale language model.

[0088] In one example, after thinking, the AI ​​agent can determine whether the execution information corresponding to the task includes memory information, tool information, or both memory information and tool information. If the AI ​​agent determines that the execution information includes tool information, it can further determine which appropriate tool information is needed to process the task.

[0089] In this way, when the AI ​​agent faces a complex problem (e.g., it cannot get the answer in a short time), it can decompose the complex problem and solve the complex problem through a series of tasks or subtasks. That is, the AI ​​agent can have the ability to decompose a complex problem. Furthermore, the AI ​​agent can determine the execution information suitable for this series of tasks through thinking, and thus comprehensively plan the processing of this series of tasks. Furthermore, the degree of intelligence of the AI ​​agent can be further improved.

[0090] In some embodiments, in response to determining that the task does not need to be processed based on the execution information, intermediate output information resulting from processing the task may be generated in step S408. This process may be performed based on the inference capabilities underlying the large-scale language model.

[0091] In one example, the intermediate output information may include feedback to a user (e.g., a response to a user) or may include feedback to an external environment (e.g., an action or response to the external environment). Note that the intermediate output information is referred to as "intermediate" output information because it is for a certain task in a series of tasks, and is distinguished from output information for input information in step S208 shown in FIG. 2 and step S308 shown in FIG. 3.

[0092] In this way, when an AI agent determines that a task can be processed without requiring additional execution information, it can directly generate intermediate output information for feedback, which makes the AI ​​agent's planning and acting capabilities more flexible and consistent.

[0093] In some embodiments, if it is determined in step S307 of the method 300 shown in FIG. 3 that the at least one processing result information does not meet the preset criteria, the method 300 may proceed to step S402 of the process 400, thereby re-determining at least one task related to processing the input information.

[0094] In this way, by redoing the task decomposition, the AI ​​agent can identify which tasks were problematic in the previous set of tasks and return to the correct tasks, avoiding going down the wrong solution path.

[0095] As previously mentioned, AI agents according to embodiments of the present disclosure may also have memory storage capabilities such that corresponding memory information is stored.

[0096] In some examples, the memory information may include working memory information, the working memory information configured to be associated with a task currently being processed.

[0097] In one example, as mentioned above, an AI agent can decompose a complex problem into a series of tasks to solve. When a task is currently being processed, the AI ​​agent needs to keep these "present" and closely related memories intact to ensure that the current task is completed with high quality.

[0098] In one example, an AI agent can generate behavior logs continuously as it interacts with a user or the external environment. For example, when an AI agent is asked a follow-up question by a user, it needs to clearly remember the specific details of the user's previous question and answer so that it can provide a coherent, logical, and smooth answer. Thus, working memory information can be stored as a behavior log.

[0099] Therefore, the working memory information mechanism can ensure that the AI ​​agent has the real-time memory content required to process the current task, and ensure that the AI ​​agent can process the current task.

[0100] In some embodiments, the working memory information is configured to associate with a processing state of each task in said at least one task.

[0101] In one example, when an AI agent is performing a sequence of tasks, it must explicitly remember which tasks have already been performed and which have not been performed in order to perform all the sequence of tasks completely. Therefore, the AI ​​agent must also maintain memory related to the processing state of the tasks.

[0102] Therefore, the working memory information mechanism can also ensure the consistency and completeness of the AI ​​agent in solving complex problems, ensuring that the AI ​​agent can solve complex problems.

[0103] In some embodiments, the storage information may include event storage information. The event storage information may be configured to associate with historical events.

[0104] In one example, event memory information may be generated by an AI agent by abstracting working memory information related to a series of "events" (e.g., after the AI ​​agent completes several sequential tasks or completes a topical discussion with a user). An event may include a cause, a history, and an effect. Thus, the event memory information may include a digest of the working memory information associated with the AI ​​agent's events.

[0105] In one example, event memory information can be stored in a vector database for a long time. When an AI agent thinks it needs to remember a certain historical event, it can find the related historical event by invoking a memory retriever and use it to process the task. For example, an event can be encoded as a query vector, and at query time, the events queried by the user can be obtained by calculating the similarity between the query vector and the vectors in the vector database.

[0106] Therefore, the event memory information mechanism continues to remember the tasks that the AI ​​agent has already completed, thereby enabling the AI ​​agent to form experience regarding them, further enhancing the degree of intelligence.

[0107] In some embodiments, the storage information may include character storage information that may be configured to associate with a scene to which the method of the embodiments of the present disclosure is applied such that the generated output information is stylized for that scene.

[0108] In one example, an AI agent can have different "human settings" under different application scenarios. The character memory information can determine the AI ​​agent's perception of its "self" and affect how it behaves and communicates in different scenarios. For example, in an e-commerce recommendation application scenario, the AI ​​agent's human setting can be generous and lively. In an emotional attendant application scenario, the AI ​​agent's human setting can be empathetic and kind. In a legal assistance application scenario, the AI ​​agent's human setting can be serious and professional. Accordingly, the AI ​​agent can give a response to the user based on these different "human settings" so that the output information it sends back to the user fits the current style.

[0109] Therefore, the character memory information mechanism not only enables the AI ​​agent to adjust its expression form according to different application scenarios, but also to better meet the needs of users, thus establishing a deeper connection with users.

[0110] In some examples, the memory information can include cognitive memory information that can be configured to be associated with a user applying a method of an example of the present disclosure such that the generated output information is tailored to a user image of the user.

[0111] In this example, if there are enough accumulated events, the AI ​​agent can abstract high-level cognition of the environment and the user from them, and can even form the AI ​​agent's unique personality and emotions, forming cognitive memory information. For example, in the process of an AI agent interacting with a user, if multiple events all show that the user has a clear strong interest and preference in a certain topic, the AI ​​agent can extract a unique user image from them, form a long-term memory, and better satisfy the user's interests and habits in subsequent interactions, thereby becoming more and more "understanding" the user.

[0112] Therefore, the cognitive memory information mechanism can encourage AI agents to fully utilize their inductive and inference abilities, further enhancing their level of intelligence.

[0113] As previously mentioned, an AI agent according to an embodiment of the present disclosure may also have computational capabilities, which may be realized through the use of tool information.

[0114] In some embodiments, the tool information includes at least one of a plug-in tool, a function tool, an interface tool, or a model tool.

[0115] In one example, a calculator may be a type of plug-in tool, a search engine may be a type of interface tool, the python language may be a type of function tool, and a large-scale language model may be a type of model tool.

[0116] In one example, an AI agent itself can be a model tool that is invoked by other AI agents, forming a multi-agent collaborative topology to generate collective intelligence.

[0117] Therefore, the AI ​​agent can have computing power by utilizing various types of tool information, which further improves the intelligence level of the AI ​​agent.

[0118] In some embodiments, the tool information may include a unified tool description specification.

[0119] In one example, a tool description specification prescribes how developers and maintainers should define the functionality, usage, and return format of tool information.

[0120] By agreeing on a unified tool description specification for various types of tool information, the occurrence of invocation errors or invocation failure problems can be avoided, thereby providing efficient and stable communication for the coupling of the control capabilities and computational capabilities of AI agents.

[0121] In some embodiments, the artificial intelligence based information processing method of the embodiments of the present disclosure (method 200 shown in FIG. 2 or method 300 shown in FIG. 3) can be performed based on the inference capabilities of a large-scale language model.

[0122] In one example, the steps of determining execution information related to the processing of input information in steps S204, S304 (i.e., planning ability), the steps of using the execution information to obtain at least one processing result information corresponding to the processing of input information in steps S206, S306 (i.e., action ability), the step of determining whether the at least one processing result information meets a preset criterion in step S307 (i.e., evaluation ability), and the step of modifying a previous process using the execution information to obtain a new result when the AI ​​agent determines that at least one memory information search result and / or at least one tool information call result is not as expected (i.e., rethinking ability) can be realized based on inference underlying a large-scale language model.

[0123] In one example, steps S402, S404, and S406 in the process 400 shown in FIG. 4 may be performed based on the inference underlying a large-scale language model.

[0124] Therefore, when performing an artificial intelligence-based information processing method, the AI ​​agent according to the embodiments of the present disclosure can further release the machine intelligence inherent in the large-scale language model based on the language understanding and generation capabilities of the large-scale language model, thereby achieving more complex decision-making and solving more complex problems.

[0125] Additionally, embodiments of the present disclosure provide an AI agent configured to execute the artificial intelligence-based information processing method described above.

[0126] The agent mechanism of the AI ​​agent according to the embodiment of the present invention fully covers the capability definitions recognized by the industry for AI agents, and the internal functions of the modules are highly unified, the division of responsibilities between modules is clear, and the communication mechanism is simple and effective.

[0127] Furthermore, the agent mechanism of the AI ​​agent according to the embodiments of the present disclosure is not designed to depend on a specific application scenario, but is a general-purpose agent mechanism suitable for a variety of environments, and therefore has a wide range of application scenarios, is highly flexible, and allows for advanced customization.

[0128] At the same time, the agent mechanism of the AI ​​agent according to the embodiment of the present disclosure improves the upper limit of the AI ​​agent's capabilities by incorporating evaluation and reconsideration capabilities into the control capabilities, and also presents the concept of cognitive memory to fully utilize the inductive reasoning capabilities of the AI ​​agent.

[0129] FIG. 5 illustrates a schematic diagram of an AI agent 500 according to an embodiment of the present disclosure.

[0130] In an embodiment of the present disclosure, as shown in FIG. 5, inspired by the von Neumann structure in modern computer theory, the AI ​​agent 500 includes five core modules: an input module 510, a control module 520, a memory module 530, a computation module 540, and an output module 550.

[0131] The input module 510 receives or recognizes information, such as queries, requests, commands, signals, or data from the outside (e.g., a user or the external environment), and converts them into a format that can be understood and processed by the AI ​​agent 500. The input module 510 is the main link through which the AI ​​agent 500 interacts with the outside world, and enables the AI ​​agent 500 to efficiently and accurately obtain necessary "sensory" information from the outside world and respond to such information.

[0132] In one example, the input module 510 may perform step S202 of the method 200 shown in FIG. 2 or step S302 of the method 300 shown in FIG.

[0133] The control module 520 is the core support for the AI ​​agent 500's ability to handle complex tasks. The control function of the control module 520 has four aspects: planning ability, action ability, evaluation ability, and reconsideration ability.

[0134] In one example, taking the method 200 shown in FIG. 2 and the method 300 shown in FIG. 3 as an example, the planning ability can correspond to step S204 or step S304. The action ability can correspond to step S206 or step S306. The evaluation ability can correspond to step S307. The reconsideration ability can correspond to the step of returning from step S307 to step S306 in the method 300 and reusing the execution information to obtain at least one processing result information corresponding to the processing of the input information when it is determined that at least one processing result information does not meet the preset criterion.

[0135] In one example, during operation, the control module 520 continually interacts with the storage module 530, the computing module 540, and / or the output module 550. However, in an embodiment of the present disclosure, the control module 520 initiates communication to the storage module 530, the computing module 540, and / or the output module 550 as a single initiator, and there is no communication coupling between the storage module 530, the computing module 540, and the output module 550.

[0136] In one example, the performance of the control module 520 is closely related to the large-scale language model on which the AI ​​agent 500 is based. To maximize the capabilities of the large-scale language model, the internal structure of the control module 520 can be designed to be highly configurable and extensible, thereby meeting different kinds of tasks and needs in real-world scenes.

[0137] The memory module 530 may be responsible for storing information such as historical interactions, event streams, etc. The character memory information, working memory information, event memory information, and cognitive memory information described above may be included within the memory module 530.

[0138] In one example, after the AI ​​agent 500 obtains input information, it can determine execution information related to the processing of the corresponding input information. If the execution information includes memory information, the AI ​​agent 500 can retrieve the relevant memory information from the memory module 530 and feed it back to the control module 520. Then, the control module 520 uses the fed back memory information to obtain at least one processing result information corresponding to the processing of the input information, and transmits the at least one processing result information to the output module 550.

[0139] The computation module 540 can be considered as a predefined tool library. As previously described, plug-in tools, function tools, interface tools, and model tools can be included in the computation module 540.

[0140] In one example, when the AI ​​agent 500 determines that the execution information includes tool information, it can call up the relevant tool information from the calculation module 540 and feed it back to the control module 520. Then, the control module 520 uses the fed back tool information to obtain at least one processing result information corresponding to the processing of the input information, and passes the at least one processing result information to the output module 550. As can be seen, large-scale language models have excellent language understanding and generation capabilities, but just like humans, there are limited tasks that they can solve without the help of any tools. When the AI ​​agent 500 is given the ability to call tools, it can realize the tasks of mathematical calculations by a computer, data analysis by Python, and weather forecasting by a search engine.

[0141] In one example, the output module 550 may perform step S208 in the method 200 shown in FIG. 2 or step S308 in the method 300 shown in FIG.

[0142] The AI ​​agent 500 according to the embodiment of the present disclosure can simply and effectively improve the level of intelligence and increase flexibility and versatility.

[0143] Additionally, the embodiments of the present disclosure further provide an information processing device based on artificial intelligence.

[0144] FIG. 6 is a block diagram illustrating a configuration of an information processing device 600 based on artificial intelligence according to an embodiment of the present disclosure.

[0145] As shown in FIG. 6, the apparatus 600 includes an input unit 602, a planning unit 604, an action unit 606, and an output unit 608.

[0146] The input unit 602 is configured to obtain input information to be processed.

[0147] The planning unit 604 is configured to determine execution information related to processing of the input information, the execution information including at least one of stored information to be retrieved or tool information to be invoked.

[0148] The behavior unit 606 is configured to utilize the execution information to obtain at least one processing result information corresponding to the processing of the input information.

[0149] The output unit 608 is configured to integrate at least one processing result information to generate output information for feedback.

[0150] The operations of the input unit 602, the planning unit 604, the action unit 606, and the output unit 608 may correspond to steps S202, S204, S206, and S208 shown in Fig. 2, respectively. Therefore, the details of each aspect will not be repeated here.

[0151] FIG. 7 is a block diagram illustrating a configuration of an artificial intelligence-based information processing device 700 according to another embodiment of the present disclosure.

[0152] 7, the apparatus 700 may include an input unit 702, a planning unit 704, an action unit 706, and an output unit 708. The operations of the input unit 702, the planning unit 704, the action unit 706, and the output unit 708 may be the same as those of the input unit 602, the planning unit 604, the action unit 606, and the output unit 608 shown in FIG.

[0153] In some embodiments, the apparatus 700 may further include an evaluation unit 705 configured to determine whether the at least one processing result information meets a preset criterion, and a reconsideration unit 707 configured to instruct the action unit 706 to reuse the execution information to obtain at least one processing result information corresponding to the processing of the input information, in response to determining that the at least one processing result information does not meet the preset criterion.

[0154] In some embodiments, the input information may be multimodal. The input unit 702 may include a conversion unit 7020 configured to convert the multimodal input information into a unified information format.

[0155] In some examples, the planning unit 704 may include a decomposition unit 7040 configured to determine, based on the input information, at least one task associated with processing the input information, and a thinking unit 7042 configured to determine, for each task of the at least one task, whether the task needs to be processed based on the execution information, and to determine execution information corresponding to the task in response to determining that the task needs to be processed based on the execution information.

[0156] In some embodiments, the thinking unit 7042 may be further configured to generate intermediate output information obtained by processing the task in response to determining that there is no need to process the task based on the execution information.

[0157] In some embodiments, the reconsideration unit 707 may be further configured to instruct the decomposition unit 7040 to determine at least one task related to processing the input information based on the input information anew in response to determining that at least one processing result information does not meet a preset criterion.

[0158] In some embodiments, the memory information may include working memory information, the working memory information configured to be associated with a task currently being processed.

[0159] In some embodiments, the working memory information is further configured to be associated with a processing state of each task in the at least one task.

[0160] In some embodiments, the storage information may include event storage information, the event storage information configured to associate with historical events.

[0161] In some embodiments, the storage information may include character storage information configured to be associated with a scene to which the apparatus 700 is applied, such that the generated output information matches a style in the scene.

[0162] In some embodiments, the memory information may include cognitive memory information, which is configured to be associated with a user applying the device 700 and align the generated output information with a user image of the user.

[0163] In some embodiments, the tool information includes at least one of a plug-in tool, a function tool, an interface tool, or a model tool.

[0164] In some embodiments, the tool information includes a unified tool description specification.

[0165] In some embodiments, the device 700 is implemented based on the inference capabilities of large-scale language models.

[0166] According to an embodiment of the present disclosure, there is further provided an electronic device, comprising at least one processor and a memory communicatively connected to the at least one processor, the memory storing instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the above-mentioned method.

[0167] According to an embodiment of the present disclosure, there is further provided a non-transitory computer readable storage medium having computer instructions stored thereon, the computer instructions being used to cause a computer to perform the above-mentioned method.

[0168] According to an embodiment of the present disclosure, there is further provided a computer program product including a computer program, which, when executed by a processor, implements the above-mentioned method.

[0169] Next, referring to FIG. 8, a configuration block diagram of an electronic device 800 that can function as a server or client of the present disclosure will be described, which is an example of a hardware device applicable to each embodiment of the present disclosure. The electronic device represents various forms of digital electronic computers, such as laptop computers, desktop computers, stages, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may further represent various forms of mobile devices, such as personal digital processing, mobile phones, intelligent phones, wearable devices, and other similar computing devices. The components, their connections, and their functions shown herein are merely exemplary and do not limit the implementation of the present disclosure as described and / or claimed herein.

[0170] 8, the electronic device 800 includes a computing unit 801 that can perform various appropriate operations and processes according to a computer program stored in a read-only memory (ROM) 802 or loaded from a storage unit 808 into a random access memory (RAM) 803. The RAM 803 may also store various programs and data required for the operation of the electronic device 800. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0171] A number of components in the electronic device 800, including an input unit 806, an output unit 807, a storage unit 808, and a communication unit 809, are connected to the I / O interface 805. The input unit 806 may be any type of device capable of inputting information to the electronic device 800, and the input unit 806 may receive input numeric or character information and generate key signal inputs for user settings and / or function control of the electronic device, including but not limited to a mouse, a keyboard, a touch screen, a track board, a track ball, a control lever, a microphone, and / or a remote control. The output unit 807 may be any type of device capable of presenting information, including but not limited to a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 808 may include but is not limited to a magnetic disk, an optical disk. The communications unit 809 enables the electronic device 800 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunications networks, and may include, but is not limited to, a modem, a network card, an infrared communications device, a wireless communications transceiver, and / or a chipset (e.g., a Bluetooth device, an 802.11 device, a WiFi device, a WiMax device, a cellular communications device, and / or the like).

[0172] The computing unit 801 may be various general-purpose and / or special-purpose processing components having processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that execute machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 executes various methods, processes, and / or processes described above. For example, in some embodiments, the methods may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 808. In some embodiments, some or all of the computer program may be loaded and / or installed in the electronic device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, it may perform one or more steps of the methods described above. Alternatively, in other embodiments, the computing unit 801 may be configured to execute the methods described above in any other suitable manner (e.g., by firmware).

[0173] Various embodiments of the systems and techniques described herein may be implemented in digital electronic circuitry systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may be embodied in one or more computer programs that may be executed and / or interpreted by a programmable system that includes at least one programmable processor, which may be a special purpose or general purpose programmable processor, and may receive data and instructions from a storage system, at least one input device, and at least one output device, and may transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0174] The program code implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, a special purpose computer or other programmable data processing apparatus, so that when executed by the processor or controller, the program code performs the functions / operations specified in the flow charts and / or block diagrams. The program code may be entirely executed on a machine, partly executed on a machine, partly executed on a machine as a separate software package and partly executed on a remote machine, or entirely executed on a remote machine or server.

[0175] In the context of the present disclosure, a machine-readable medium may be a tangible medium, which may include or store a program for use in or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the above. More specific examples of machine-readable storage media include an electrical connection with one or more leads, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0176] To provide for interaction with a user, a computer may implement the systems and techniques described herein and include a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user, and a keyboard and pointing device (e.g., a mouse or trackball) through which a user may provide input to the computer. Other types of devices may also be provided for providing interaction with a user, for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or haptic feedback) and may receive input from the user in any form (including sound input, speech input, or tactile input).

[0177] The systems and techniques described herein may be implemented in a computing system including a backstage component (e.g., as a data server), a middleware component (e.g., an application server), a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with the system or technique embodiment), or any combination of the backstage component, middleware component, or front-end component. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), the Internet, and a blockchain network.

[0178] The computer system may include a client side and a server. The client side and the server are generally remote from each other and usually interact with each other via a communication network. The relationship between the client side and the server is generated by running computer programs having a client side-server relationship with each other on corresponding computers. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0179] It should be understood that the steps may be reordered, added, or removed using the various forms of flow described above. For example, the steps described in this disclosure may be performed in parallel, sequentially, or in a different order, and the present disclosure is not limited thereto as long as the technical solutions disclosed in this disclosure can achieve the desired results.

[0180] In the technical solution disclosed herein, the collection, storage, use, processing, transmission, provision and disclosure of relevant user personal information shall all comply with the provisions of relevant laws and regulations and shall not violate public order and morals.

[0181] Although the embodiments or examples of the present disclosure have been described with reference to the drawings, it should be understood that the above methods, systems, and devices are merely exemplary embodiments or examples, and the scope of the present invention is not limited by these embodiments or examples, but only by the scope of the claims and their equivalents after grant. Various elements of the embodiments or examples may be omitted or replaced by their equivalent elements. In addition, each step may be performed in a different order from the order described in this disclosure. Furthermore, various elements of the embodiments or examples may be combined in various ways. It is important to note that with the evolution of technology, many elements described herein may be replaced by equivalent elements that appear later in this disclosure.

Claims

1. 1. An information processing method based on artificial intelligence, comprising: Obtaining input information to be processed; determining execution information associated with processing the input information, the execution information including at least one of stored information to be retrieved or tool information to be invoked; obtaining at least one piece of processing result information corresponding to processing of the input information using the execution information; and synthesizing the at least one processing result information to generate output information for feedback.

2. determining whether the at least one processing result information meets a preset criterion; 2. The method of claim 1, further comprising: in response to determining that the at least one processing result information does not meet the preset criterion, reusing the execution information to obtain at least one processing result information corresponding to processing of the input information.

3. 3. The method of claim 1 or 2, wherein the input information is multi-modal, and wherein obtaining the input information to be processed comprises converting the multi-modal input information into a unified information format.

4. determining execution information related to processing the input information includes: determining, based on the input information, at least one task associated with processing the input information; For each task of the at least one task, determining whether the task needs to be processed based on the execution information; 3. The method of claim 2, further comprising: determining the performance information corresponding to the task in response to determining that the task needs to be processed based on the performance information.

5. The method of claim 4 , further comprising: generating intermediate output information resulting from processing the task in response to determining that the task does not need to be processed based on the execution information.

6. 5. The method of claim 4, further comprising, in response to determining that the at least one processing result information does not meet the preset criteria, redefining at least one task associated with processing the input information.

7. The method of claim 4 , wherein the stored information comprises working memory information, the working memory information being configured to be associated with the task currently being processed.

8. The method of claim 7 , wherein the working memory information is further configured to be associated with a processing state of each task in the at least one task.

9. The method of claim 1 or 2, wherein the stored information comprises event stored information, the event stored information being configured to be associated with historical events.

10. The method of claim 1 or 2, wherein the storage information includes character storage information, the character storage information being configured to be associated with a scene to which the method is applied, and the generated output information is adapted to match a style in the scene.

11. The method of claim 1 or 2, wherein the memory information comprises cognitive memory information, the cognitive memory information being configured to be associated with a user applying the method, and the generated output information is tailored to a user image of the user.

12. The method of claim 1 or 2, wherein the tool information includes at least one of a plug-in tool, a function tool, an interface tool, or a model tool.

13. The method of claim 12 , wherein the tool information comprises a unified tool description specification.

14. The method according to claim 1 or 2, wherein the method is performed based on the inference capabilities of a large-scale language model.

15. An artificial intelligence agent arranged to carry out the method according to any one of claims 1 to 14.

16. An information processing device based on artificial intelligence, an input unit configured to obtain input information to be processed; a planning unit configured to determine execution information associated with processing the input information, the execution information including at least one of stored information to be retrieved or tool information to be recalled; a behavior unit configured to utilize the execution information to obtain at least one processing result information corresponding to processing the input information; and an output unit configured to integrate the at least one processing result information to generate output information for feedback.

17. an evaluation unit configured to determine whether the at least one processing result information meets a preset criterion; and a reconsideration unit configured to instruct the behavior unit to reuse the execution information to obtain at least one processing result information corresponding to processing of the input information in response to determining that the at least one processing result information does not meet the preset criterion.

18. 18. The apparatus according to claim 16 or 17, wherein the input information is multimodal in nature, and the input unit comprises a conversion unit configured to convert the multimodal input information into a unified information format.

19. The planning unit comprises: a decomposition unit configured to determine, based on the input information, at least one task related to processing the input information; For each task of the at least one task, determining whether the task needs to be processed based on the execution information; and a thinking unit configured to determine the execution information corresponding to the task in response to determining that the task needs to be processed based on the execution information.

20. The thought unit further comprises:

20. The apparatus of claim 19, configured to generate intermediate output information resulting from processing the task in response to determining that the task does not need to be processed based on the execution information.

21. The apparatus of claim 19, wherein the reconsideration unit is further configured to instruct the decomposition unit to determine at least one task related to processing the input information based on the input information anew in response to determining that the at least one processing result information does not meet the preset criterion.

22. 20. The apparatus of claim 19, wherein the stored information comprises working memory information, the working memory information being configured to be associated with the task currently being processed.

23. 23. The apparatus of claim 22, wherein the working memory information is further configured to be associated with a processing state of each task in the at least one task.

24. 18. Apparatus according to claim 16 or 17, wherein the stored information comprises event stored information, the event stored information being arranged to be associated with historical events.

25. 18. An apparatus as claimed in claim 16 or 17, wherein the storage information includes character storage information, the character storage information being arranged to be associated with a scene to which the apparatus is applied, and wherein the generated output information is adapted to match a style in the scene.

26. 18. The device of claim 16 or 17, wherein the memory information comprises cognitive memory information, the cognitive memory information being configured to be associated with a user applying the device, and wherein the generated output information is tailored to a user image of the user.

27. The apparatus of claim 16 or 17, wherein the tool information includes at least one of a plug-in tool, a function tool, an interface tool, or a model tool.

28. 30. The apparatus of claim 27, wherein the tool information comprises a unified tool description specification.

29. 18. The device according to claim 16 or 17, wherein the device is implemented based on the inference capabilities of a large-scale language model.

30. An electronic device, At least one processor; a memory communicatively coupled to the at least one processor; An electronic device, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform a method according to any one of claims 1 to 14.

31. A non-transitory computer readable storage medium having stored thereon computer instructions, the computer instructions being used to cause a computer to perform the method of any one of claims 1 to 14.

32. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 14.

Images