System

Abstract

A system is provided.SOLUTION: A system, comprising: means for registering AI models; means for scheduling a match; means for notifying participating AI models of a challenge; means for receiving answers generated by the AI models based on the challenge; means for scoring the generated answers based on evaluation criteria; and means for recording and publishing match results.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] In recent years, the development and use of generative AI has progressed rapidly, but there is still a lack of an environment in which the performance of AI models can be objectively evaluated and developers can compete with each other. Furthermore, there are limited ways for users to directly observe and deepen their understanding of the performance of each AI model. Furthermore, there is a lack of opportunities for AI developers to hone and optimize their skills by competing with other AI models. [Means for solving the problem]

[0005] To solve the above-mentioned problems, the present invention provides the following means. First, a system including means for registering AI models is constructed. Next, a means for scheduling matches is provided, and a means for notifying participating AI models of challenges is provided. Furthermore, a system including means for receiving answers generated by AI models based on challenges and scoring them based on evaluation criteria is provided. Finally, a means for recording and publishing match results is provided, thereby objectively evaluating the performance of AI models and providing an environment that can be utilized by developers and users. This allows AI developers to improve their skills by competing with other AI models, and users to directly observe and deepen their understanding of the performance of each AI model.

[0006] An "AI model" is an artificial intelligence algorithm or system designed to perform a specific task.

[0007] "Registration" is the process of formally submitting an AI model to the system and assigning it a unique ID.

[0008] A "match" is an event in which multiple AI models compete on a specific task and their performance is compared and evaluated.

[0009] "Scheduling" is the process of determining the date and time when a match will be played and which AI models will participate.

[0010] A "problem" is a definition of the problem or task that an AI model must address.

[0011] "Notification" is the act of sending information about a set task to an AI model.

[0012] An "answer" is the output result generated by an AI model for a task.

[0013] "Receiving" is the process by which the system receives the answer sent from the AI ​​model.

[0014] "Evaluation criteria" are measures or indicators used to evaluate the answers generated by an AI model.

[0015] "Scoring" is the process of assigning a score to an AI model's answer based on evaluation criteria.

[0016] "Recording" refers to the act of saving the results of a match in a database or the like.

[0017] "Publication" means the act of making match results accessible to developers and users.

[0018] A "system" is a structure in which multiple components work together to provide a specific function.

[0019] "Developers" are engineers who design and implement AI models.

[0020] A "user" is someone who uses the system to observe and understand the performance of the AI ​​model. [Brief explanation of the drawings]

[0021] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7]FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0022] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0023] First, the terms used in the following description will be explained.

[0024] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0025] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0026] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0027] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0028] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0029] [First embodiment]

[0030] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0031] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0032] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0033] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0034] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0035] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0036] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0037] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0038] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0039] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0040] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0041] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0042] This system provides an environment in which generative AIs tackle problems and compete with each other to find the best solutions. The system mainly consists of a server, a terminal (for developers), and a user. The specific functions and operations of each component are explained below.

[0043] server

[0044] 1. Registering an AI model

[0045] The server receives the AI ​​model uploaded from the developer's device. After receiving it, it assigns a unique ID to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. It also performs a security check to ensure that the model does not contain any malicious programs.

[0046] 2. Match Scheduling

[0047] The server schedules the next match based on a pre-defined schedule, selects which AI models will participate, and records match details (e.g., task content, evaluation criteria) in a database.

[0048] 3. Match execution

[0049] When the match starts, the server notifies the AI ​​models of the challenge. The challenge notification includes details such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on the evaluation criteria.

[0050] 4. Recording and publishing match results

[0051] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[0052] Terminal (for developers)

[0053] 1. Submitting an AI model

[0054] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[0055] 2. Check the match results

[0056] Developers can view their model's scores and detailed feedback through a match results interface, which allows them to improve their AI model for the next match.

[0057] User

[0058] 1. Watching a game

[0059] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[0060] 2. Analysis of results

[0061] After the match, users can analyze the results in detail, understanding how each AI model solved the task and how the score based on the evaluation criteria was determined.

[0062] Specific examples

[0063] Example 1: Text generation task

[0064] The server sets the task of "generating a poem." Developer A submits AI model A, and developer B submits AI model B. When the match begins, the server displays a prompt to "generate a poem with an autumn landscape theme." Each AI model generates a poem and submits it to the server. The server evaluates the generated poems and assigns them scores based on criteria such as literary merit and grammatical accuracy.

[0065] Example 2: Image classification task

[0066] The server sets the task of "classifying cat images." Developer C submits AI model C, and developer D submits AI model D. When the match begins, the server notifies the AI ​​model of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server scores the AI ​​models based on accuracy rate and classification speed.

[0067] In this way, the AI ​​Battlefield system provides an environment in which developers can compete against other AI models and a means for users to observe the progress of AI technology in real time.

[0068] The processing flow will be explained below.

[0069] Step 1:

[0070] Uploading an AI model

[0071] Device: Developers upload the AI ​​model from their own device via the web interface, select the model from the file selection screen, and press the send button.

[0072] Step 2:

[0073] Security checks for AI models

[0074] Server: Receives the uploaded AI model and performs security checks to ensure it does not contain malicious code or viruses.

[0075] Step 3:

[0076] Registering an AI model

[0077] Server: After security checks are complete, it assigns a unique ID to the AI ​​model and stores the model and its metadata (developer name, model name, version, etc.) in a database.

[0078] Step 4:

[0079] Match Scheduling

[0080] Server: Schedules the next match based on a pre-set schedule. At this time, it selects the participating AI models and records the match details (task content, evaluation criteria) in a database.

[0081] Step 5:

[0082] Assignment notifications

[0083] Server: When the match start time arrives, the server notifies the participating AI models of the challenge, including details such as the prompt and the dataset URL.

[0084] Step 6:

[0085] Generate answers

[0086] Terminal: Each participating AI model generates an answer to the task and sends the result to the server. The answer generation must be completed within a specified time.

[0087] Step 7:

[0088] Receiving answers

[0089] Server: Receives the answers sent from each AI model and stores them in a database. Monitors the reception status and confirms that all answers have been received.

[0090] Step 8:

[0091] Scoring answers

[0092] Server: Scores the received answers based on evaluation criteria, such as originality, accuracy, and efficiency. Stores the scoring results in a database.

[0093] Step 9:

[0094] Recording match results

[0095] Server: Generates match results based on the scoring results and records them in a database, including the scores of each AI model, generated answers, and evaluation comments.

[0096] Step 10:

[0097] Publication of match results

[0098] Server: The server publishes the match results on a web interface accessible to developers and users. The results page displays the scores and evaluation comments for each AI model.

[0099] Step 11:

[0100] Check the match results

[0101] Device: Developers can view match results, scores, and detailed feedback from their own devices through a web interface, which can be used to improve the game for the next time.

[0102] Step 12:

[0103] Watching a game

[0104] Users: Users can watch the matches in real time and observe the performance of each AI model, and after the match, they can also perform detailed analysis of the results.

[0105] Example 1

[0106] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0107] Conventional competition systems using AI models lacked an efficient workflow, from model registration to running matches, publishing results, and analyzing them. Furthermore, they often lacked important features such as security checks and detailed feedback, resulting in issues of low usability and reliability for AI developers. There was a need for an effective system that could solve these issues, promote competition between AI models, and present technological progress to users in real time.

[0108] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0109] In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for performing security checks, means for assigning unique identifiers to AI models, means for users to watch matches in real time through a web interface, means for analyzing match results in detail, means for providing detailed feedback on generated answers, and means for providing developers with information to improve their AI models. This realizes an environment in which a series of processes, from registering AI models to detailed analysis of match results, can be executed efficiently and reliably, and makes it possible to provide AI developers with useful information.

[0110] An "AI model" is software designed to perform a specific task using artificial intelligence techniques.

[0111] A "server" refers to a computer system that provides services to other devices over a network.

[0112] "Device" means an electronic device used by a user or developer to access the Service.

[0113] "User" means any person or organization that uses the system or service.

[0114] A "challenge" is a task or problem that an AI model is given to solve.

[0115] An "answer" is the result or response that an AI model generates based on a task.

[0116] "Evaluation criteria" refers to the criteria or scale used to score the generated answers.

[0117] "Scoring" refers to assigning points to generated answers based on evaluation criteria.

[0118] A "database" refers to a system for efficiently storing, managing, and retrieving information.

[0119] A "security check" is an inspection to ensure that a system or software does not contain malicious programs.

[0120] A "unique identifier" is a number or code assigned to an element to uniquely identify it from all other elements.

[0121] "Web interface" refers to a web page or application that can be accessed through a browser.

[0122] "Feedback" refers to evaluations and comments on the AI ​​model's answers, which can be used to improve the model.

[0123] "Developers" refers to the engineers and programmers who create AI models and submit them to the system.

[0124] "Real-time" means that an event is processed and displayed immediately, with almost no delay, after it occurs.

[0125] "Registration" refers to the act of adding new information or models to the system.

[0126] "Scheduling" refers to setting a timetable for the execution of a particular event or task.

[0127] This invention is a system that provides an environment in which generative AI models tackle problems and compete with each other to find the best solutions. This system is mainly composed of a server, a terminal (for developers), and a user.

[0128] server

[0129] The server plays a central role in managing and operating the entire system using multiple pieces of hardware and software. The main functions of the server are as follows:

[0130] Registering an AI model

[0131] The server receives the AI ​​model uploaded from the developer's device and assigns it a unique identifier. The received AI model and related information (developer name, model name, version, etc.) are stored in a database. A security check is performed to ensure that the model does not contain any malicious programs.

[0132] Match Scheduling

[0133] The server schedules the next match based on a pre-defined schedule, selects the participating AI models, and records match details (e.g., task content, evaluation criteria) in a database.

[0134] Match execution

[0135] When the match starts, the server notifies the AI ​​models of the challenge. This challenge notification includes detailed information such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on the evaluation criteria.

[0136] Recording and publishing match results

[0137] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[0138] As a concrete example, the server can set a challenge to "generate a poem." For example, developer A submits AI model A, and developer B submits AI model B. When the match starts, the server notifies the participants with a prompt to "generate a poem on the theme of autumn scenery." Each AI model generates a poem and submits it to the server. The server evaluates the generated poems and assigns them scores based on criteria such as literary merit and grammatical accuracy.

[0139] Terminal (for developers)

[0140] The terminals are used by developers to submit AI models and check match results. Developers can use their own terminals to submit AI models to the server through a web interface and check match results.

[0141] Submitting an AI model

[0142] Developers can submit AI models to the server via a web interface using their own devices, and can enter model summary and version information when submitting.

[0143] Check the match results

[0144] Developers can view match results through a web interface, view their model's scores and detailed feedback, and improve their AI models for the next match.

[0145] User

[0146] Users use the system primarily to watch games and analyze results.

[0147] Watching a game

[0148] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[0149] Analyzing the results

[0150] After the match, users can analyze the results in detail, understanding how each AI model solved the task and how the score based on the evaluation criteria was determined.

[0151] As a concrete example, if the server sets the task of "classifying images of cats," developer C submits AI model C, and developer D submits AI model D. When the time for the match to start arrives, the server notifies it of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server then scores them based on accuracy rate and classification speed.

[0152] Examples of prompts include "Generate a poem on the theme of autumn scenery" and "Determine whether the image below is a cat."

[0153] The system promotes competition between AI models and provides an effective environment where developers and users can obtain information tailored to their respective goals.

[0154] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0155] Step 1: Registering an AI model

[0156] The server receives the AI ​​model from the developer's device. Specifically, the developer uploads the AI ​​model file through a web interface, and the server receives it. The input data is the AI ​​model file, and the output is the model's unique identifier and registration results. After receiving it, the server assigns a unique identifier to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. In addition, it performs security checks to ensure the AI ​​model does not contain malicious programs. Security checks include virus scans and static code analysis.

[0157] Step 2: Scheduling matches

[0158] The server schedules the next match based on a pre-set schedule. The input data is a list of AI models in the database and schedule information, and the output is detailed match information. Specifically, the server selects models to participate in the match from the list of AI models stored in the database. Past performance and random selection are used as criteria. Match details include task content and evaluation criteria, and this information is recorded in the database.

[0159] Step 3: Assignment notification

[0160] When the match starts, the server notifies the AI ​​models of the task. The input data is the task content and a list of participating AI models, and the output is the notification result. The server notifies each AI model of detailed information such as the prompt text and the URL of the dataset. Specifically, the server communicates the task to each model using an API call.

[0161] Step 4: Generate and submit answers

[0162] Each AI model generates an answer to the task and sends it to the server. The input data is the task notification content, and the output is the generated answer. Specifically, the AI ​​model generates an answer based on the received prompt. For example, it generates a poem in response to the prompt "Generate a poem on the theme of autumn scenery." The generated answer is then sent to the server.

[0163] Step 5: Receiving and evaluating answers

[0164] The server receives the answers from the AI ​​model and scores them based on evaluation criteria. The input data is the received answers, and the output is the evaluation results and score. The server scores the received answers based on evaluation criteria such as literary value and grammatical accuracy.

[0165] Step 6: Record and publish match results

[0166] The server records the match results in a database and makes them publicly available through a web interface. The input data are the evaluation results and scores, and the output is the publicly available information. The server stores the scores, generated answers, and evaluation comments of each AI model in a database and makes them publicly available through a web interface accessible to developers and users.

[0167] Step 7: Submit your AI model

[0168] Developers submit AI models through a web interface using their own devices. The input data is the AI ​​model file, its summary, and version information, and the output is the submission results. Developers select and upload the model file, enter the model summary and version information, and submit.

[0169] Step 8: Check the match results

[0170] Developers check the match results through a web interface. The input data is the match result information, and the output is the displayed score and feedback. Developers can check the scores and detailed feedback of their AI models and use them to improve their AI models for the next match.

[0171] Step 9: Watch the game

[0172] Users use an interface that allows them to watch the game in real time. The input data is game information updated in real time, and the output is the viewing experience. Users can observe live how each AI model tackles the problem and what solutions it generates.

[0173] Step 10: Analyze the results

[0174] Users can analyze the match results in detail after the match is over. The input data is the published match result information, and the output is the analysis results. Users can analyze the answers generated by each AI model and their scores based on the evaluation criteria to understand how they solved the problem. By referring to the evaluation comments, users can grasp the strengths and areas for improvement of the AI ​​model.

[0175] (Application example 1)

[0176] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0177] Previous systems utilizing generative AI models simply evaluated and compared the performance of AI models, lacking the ability to incorporate user participation and evaluation in real time. Furthermore, there were insufficient means for developers to receive detailed feedback, making it time-consuming and labor-intensive to improve AI models. Furthermore, there was a lack of platforms where users could evaluate and enjoy generated content, limiting opportunities for the general public to understand and enjoy technological advances in generative AI.

[0178] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0179] In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for users to view and evaluate generated content in real time, and means for developers to receive detailed feedback on the models. This allows the technology of generative AI models to be utilized while incorporating users' real-time evaluations, allowing developers to quickly improve their models based on detailed feedback. Furthermore, the server functions as a platform where general users can enjoy content created by generative AI, thereby promoting widespread understanding of advances in generative AI technology.

[0180] An "AI model" is an algorithm and database that uses artificial intelligence technology to generate answers and predictions for specific problems.

[0181] A "problem" is a specific problem or task that an AI model must address, and the resulting solutions are evaluated against it.

[0182] "Answer" refers to the output generated by an AI model based on a task.

[0183] "Users" are participants who use the system to view and evaluate generated content in real time.

[0184] "Developer" refers to a person or organization that develops an AI model and registers it in the system to test its performance.

[0185] "Means for notifying tasks" refers to a function that allows the server to communicate the tasks that need to be solved to the AI ​​model.

[0186] "Means for receiving answers" refers to the function for receiving answers generated from the AI ​​model on the server side.

[0187] "Means for scoring" refers to a function that assigns points to received answers based on evaluation criteria.

[0188] "Means for recording and publishing match results" refers to the ability to store the evaluation results of scored answers and display them in a form accessible to users and developers.

[0189] The "means for viewing and rating in real time" is a function that allows users to view generated content in real time and rate it on the spot.

[0190] "Means for receiving feedback" is a function that allows developers to receive detailed comments and suggestions for improvement regarding the AI ​​model's match results and user evaluations.

[0191] This invention is a system for pitting generative AI models against each other in specific challenges, allowing users to view and evaluate the generated content. Specifically, it consists of a server, a developer terminal, and a user interface.

[0192] server

[0193] 1. Registering the AI ​​model:

[0194] The server receives the AI ​​model provided by the developer, assigns a unique ID, and stores it in a database. It also performs security checks to ensure that no malicious programs are included. This process can be implemented using Python and Flask, and TensorFlow and SQLite are used to store the AI ​​model.

[0195] 2. Match Scheduling:

[0196] The server schedules matches, determines the next match time, and selects the AI ​​models that will participate. This information is recorded in a database and notifications are sent as needed. The scheduling logic is automatic, based on pre-defined times and conditions.

[0197] 3. Assignment Notification:

[0198] When the game starts, the AI ​​model is notified of the task. The task includes a prompt and the URL of the required dataset. For example, the prompt might be "Generate a poem based on an autumn landscape." This process is notified to the AI ​​model via an HTTP request.

[0199] 4. Receiving and Scoring Answers:

[0200] The answers generated by each AI model are sent to the server, where they are then scored based on evaluation criteria. Scoring evaluates the quality of the generated content and its suitability for the task. This process uses natural language processing (NLP) technology and evaluation algorithms.

[0201] 5. Recording and Publication of Match Results:

[0202] Match results are recorded in a database and made publicly accessible to users and developers, allowing users to rate the generated content and developers to receive feedback.

[0203] Developer Device

[0204] 1. Submitting an AI model:

[0205] Developers submit AI models to the server from their own devices, providing details such as the model name, version information, and developer details. Submissions are done through a web-based interface.

[0206] 2. Match result confirmation and feedback:

[0207] Developers have an interface that allows them to view match results and detailed evaluation feedback for each model, providing them with immediate information they need to improve their models.

[0208] User

[0209] 1. Real-time viewing and evaluation:

[0210] Users can watch the progress of the match in real time. The generated content is displayed in real time, and users can rate and comment on it on the spot. This function enables user-participation in evaluation.

[0211] 2. Analysis of match results:

[0212] After the match, users can analyze the evaluation results and scoring details of the generated content, helping them see how their own evaluations are reflected and understand the technological progress of the generating AI.

[0213] Specific examples

[0214] One weekend, the task is to "generate a poem with the theme of autumn scenery." Developer A's model generates a poem that "speaks of autumn, with red and yellow leaves fluttering in the air, and sunsets that color the sky. A poem that resonates with the heart, like a whisper of the wind." Meanwhile, Developer B's model generates a poem that "speaks of autumn scenery that evokes warm memories in the heart, with cold winds blowing." Each poem is evaluated by users and assigned a score.

[0215] Prompt Sentence Examples

[0216] "Generate a poem with the theme of autumn scenery. We're looking for poems that incorporate the beautiful colors, the sound of the wind, the glow of the sunset, etc."

[0217] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0218] Step 1: Registering an AI model

[0219] Input: Developers upload the AI ​​model file from their own device to the server.

[0220] Processing: The server receives the uploaded AI model, assigns a unique ID, and records related information such as the model name, developer name, and version in a database. It also performs security checks to ensure there are no malicious programs.

[0221] Output: The AI ​​model is saved in a database, along with the model's unique ID and related information.

[0222] Step 2: Scheduling the match

[0223] Input: The server plans the next match based on a predefined match schedule.

[0224] Processing: The server selects participating AI models according to the schedule and records the match date, time, task content, and evaluation criteria in a database.

[0225] Output: The next match information is saved in the database and ready to be notified.

[0226] Step 3: Issue notification

[0227] Input: As the start time of the match approaches, the server notifies the AI ​​model of the task.

[0228] Processing: The server sends each AI model a prompt and the URL of the dataset via an HTTP request. For example, the prompt might be "Generate a poem based on an autumn landscape."

[0229] Output: Each AI model receives the task information.

[0230] Step 4: Generate and submit your answers

[0231] Input: The AI ​​model generates an answer based on the received challenge.

[0232] Processing: The AI ​​model follows the prompt, uses natural language processing techniques to generate poetry or other content, and sends the answer to the server.

[0233] Output: The generated answer is sent to the server.

[0234] Step 5: Receiving and scoring answers

[0235] Input: The server receives the answers sent by each AI model.

[0236] Processing: The server collects the answers and scores them based on a set of criteria, including literary merit and grammatical accuracy.

[0237] Output: The answer and score of each AI model are recorded in a database.

[0238] Step 6: Record and publish match results

[0239] Input: After scoring is complete, the server compiles the match results.

[0240] Processing: The server records the match results in a database and exposes them through an interface, making this information accessible to users and developers.

[0241] Output: The match results are published and available for users and developers to view.

[0242] Step 7: Real-time viewing and evaluation by users

[0243] Input: Users access the generated content to watch it in real time.

[0244] Processing: The server displays the generated content to the user in real time and provides a mechanism for the user to immediately submit ratings and comments.

[0245] Output: User ratings and comments are recorded in the system.

[0246] Step 8: Detailed analysis of match results

[0247] Input: After the match is over, users and developers have access to analyze the match results.

[0248] Processing: The server displays detailed match results and feedback, and provides tools that allow users to analyze the performance of each AI model.

[0249] Output: Detailed match results and analytical information are displayed to users and developers.

[0250] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0251] This system provides an environment in which generative AIs can tackle problems and compete with each other to find the best answers, and also combines an emotion engine that recognizes the user's emotions. The system mainly consists of a server, a terminal (for developers), and a user, and each component works in cooperation. The specific functions and operations of each component are explained below.

[0252] server

[0253] 1. Registering an AI model

[0254] The server receives the AI ​​model uploaded from the developer's device. After receiving it, the server assigns a unique ID to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. It also performs security checks to ensure that the model does not contain any malicious programs.

[0255] 2. Match Scheduling

[0256] The server schedules the next match based on a pre-defined schedule, selects which AI models will participate, and records match details (task content, evaluation criteria) in a database.

[0257] 3. Match execution

[0258] When the match starts, the server notifies the AI ​​models of the challenge. This notification includes details such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on the evaluation criteria.

[0259] 4. Recording and publishing match results

[0260] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[0261] 5. Emotion Engine

[0262] The server is equipped with an emotion engine that recognizes the user's emotions. This emotion engine analyzes the user's facial expressions, tone of voice, input data, etc. while watching a game or viewing game results, and recognizes the user's emotional state.

[0263] 6. Emotion-Based Interface Adjustment

[0264] The server dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine, for example, adjusting the interface's color scheme and the timing of information presentation if the user is excited.

[0265] Terminal (for developers)

[0266] 1. Submitting an AI model

[0267] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[0268] 2. Check the match results

[0269] Developers can view their model's scores and detailed feedback through a match results interface, which allows them to improve their AI model for the next match.

[0270] User

[0271] 1. Watching a game

[0272] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[0273] 2. Analysis of results

[0274] After a match, users can analyze the results in detail, understanding how each AI model solved the problem and how the score was determined based on the evaluation criteria. Furthermore, based on the emotional data recognized by the emotion engine, analysis and feedback are provided according to the user's emotional state.

[0275] Specific examples

[0276] Example 1: Text generation task

[0277] The server sets the task of "generating a poem." Developer A submits AI model A, and developer B submits AI model B. When the match begins, the server displays a prompt to "generate a poem with an autumn landscape theme." Each AI model generates a poem and sends it to the server. The server evaluates the generated poems and assigns them scores based on criteria such as literary value and grammatical accuracy. While watching the match, if the user's emotions are high, the interface changes to a brighter color tone.

[0278] Example 2: Image classification task

[0279] The server sets the task of "classifying cat images." Developer C submits AI model C, and developer D submits AI model D. When the match begins, the server notifies the user of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server scores the users based on accuracy rate and classification speed. When checking the match results, if the user is detected as confused, the interface displays additional support information appropriate to the situation.

[0280] In this way, the AI ​​Battlefield system not only provides a competitive environment between generative AI models, but also adjusts the interface to take user emotions into account, providing a more fulfilling experience.

[0281] The processing flow will be explained below.

[0282] MODE FOR CARRYING OUT THE INVENTION

[0283] Server Operation

[0284] Step 1:

[0285] Receiving AI model uploads

[0286] Server: Receives the AI ​​model uploaded from the developer's device. After receiving it, it starts a security check.

[0287] Step 2:

[0288] Security Check

[0289] Server: Performs security checks on the received AI model to check for malicious code and viruses.

[0290] Step 3:

[0291] Registering an AI model

[0292] Server: Once the security check is complete, it assigns a unique ID to the AI ​​model and stores the model and its metadata (developer name, model name, version) in a database.

[0293] Step 4:

[0294] Match Scheduling

[0295] Server: Schedules the next match based on a pre-set schedule, selects participating AI models, and records detailed match information (task content, evaluation criteria) in a database.

[0296] Step 5:

[0297] Assignment notifications

[0298] Server: When the match start time arrives, the server notifies the participating AI models of the challenge, including details such as the prompt and the dataset URL.

[0299] Step 6:

[0300] Receiving answers

[0301] Server: Receives answers from each participating AI model and stores them in a database. Monitors the reception status and confirms that all answers have been received.

[0302] Step 7:

[0303] Scoring

[0304] Server: Scores the received answers based on criteria such as originality, accuracy, and efficiency.

[0305] Step 8:

[0306] Recording the results

[0307] Server: Generates match results based on the scoring results and records them in a database, including the scores of each AI model, generated answers, and evaluation comments.

[0308] Step 9:

[0309] Publication of match results

[0310] Server: The server publishes the match results on a web interface accessible to developers and users. The results page displays the scores and evaluation comments for each AI model.

[0311] Emotion Engine Operation

[0312] Step 10:

[0313] User Emotion Recognition

[0314] Server: Uses an emotion engine to recognize emotions from the user's facial expressions, tone of voice, and input data.

[0315] Step 11:

[0316] Interface Adjustments

[0317] Server: Dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine. For example, if the user is excited, the color scheme of the interface or the timing of information presentation is adjusted.

[0318] User Actions

[0319] Step 12:

[0320] Watching a game

[0321] Users: Watch the match in real time and observe the performance of each AI model.

[0322] Step 13:

[0323] Analyzing the results

[0324] User: After the match, the results are analyzed in detail. Based on the emotional data recognized by the emotion engine, analysis and feedback are provided according to the user's emotional state.

[0325] Device (developer) operation

[0326] Step 14:

[0327] Submitting an AI model

[0328] Device: Developers submit AI models from their own devices through a web interface, entering model overview and version information.

[0329] Step 15:

[0330] Check the match results

[0331] On-device: Developers can view their model's scores and detailed feedback through an interface that allows them to check the match results and make adjustments to improve their model for next time.

[0332] Specific examples

[0333] Example 1: Text generation task

[0334] Step 1:

[0335] Assignment settings

[0336] Server: The server sets the task of "creating a poem."

[0337] Step 2:

[0338] Submitting an AI model

[0339] Terminal: Developer A submits AI model A, and developer B submits AI model B.

[0340] Step 3:

[0341] Assignment notifications

[0342] Server: When the match start time arrives, the server will announce the prompt to "Generate a poem with an autumn landscape theme."

[0343] Step 4:

[0344] Generating and receiving answers

[0345] Terminal: Each AI model generates a poem and sends it to the server, which receives the answer and evaluates it.

[0346] Step 5:

[0347] Scoring

[0348] Server: Evaluates the generated poems, scoring them on criteria such as literary merit and grammatical accuracy.

[0349] Step 6:

[0350] emotion recognition

[0351] Server: While watching the game, the emotion engine recognizes the user's emotions.

[0352] Step 7:

[0353] Interface Adjustments

[0354] Server: If the user is emotionally excited, change the interface color to a brighter tone.

[0355] Example 2: Image classification task

[0356] Step 1:

[0357] Assignment settings

[0358] Server: The server sets the task "classify images of cats."

[0359] Step 2:

[0360] Submitting an AI model

[0361] Terminal: Developer C submits AI model C, and developer D submits AI model D.

[0362] Step 3:

[0363] Assignment notifications

[0364] Server: When the match starts, a series of cat images will be displayed.

[0365] Step 4:

[0366] Generating and receiving answers

[0367] Terminal: Each AI model classifies an image and sends the results to the server, which receives and evaluates the answers.

[0368] Step 5:

[0369] Scoring

[0370] Server: Scoring is based on accuracy rate and classification speed.

[0371] Step 6:

[0372] emotion recognition

[0373] Server: When checking the match results, the emotion engine recognizes the user's emotions.

[0374] Step 7:

[0375] Interface Adjustments

[0376] Server: If the user is detected as confused, the interface displays additional support information.

[0377] In this way, the AI ​​Battlefield system provides a competitive environment for generative AI models and adjusts the interface to take user emotions into account, providing a more fulfilling experience.

[0378] Example 2

[0379] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0380] Conventional AI model competition systems simply record and publish the results of AI model competitions, without adjusting the interface to take into account the user's emotional state. This resulted in a lack of improvement in the user experience and reduced engagement with the system. Furthermore, security checks and scheduling of AI models were not performed efficiently, leaving the system lacking reliability and operational efficiency.

[0381] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for analyzing the user's emotional state, and means for dynamically adjusting the interface based on the user's emotional state. This makes it possible to provide the results of AI model matches in a real-time and reliable manner and to improve the user experience.

[0382] An "AI model" is an artificial intelligence software system that has been trained to perform a specific task.

[0383] A "match" is an event in which multiple AI models compete against each other based on a given task and their performance is evaluated.

[0384] A "problem" is a specific problem or requirement that the AI ​​model must solve, such as text generation or image classification.

[0385] "Notification" is an action in which the server communicates details of the problem to the AI ​​model.

[0386] An "answer" is the resulting data generated by an AI model based on a task.

[0387] "Evaluation criteria" are indicators and rules for evaluating answers generated by an AI model.

[0388] "Scoring" is the process of numerically evaluating an AI model's answer based on evaluation criteria.

[0389] "Match results" are data that include the evaluation results of each AI model's answers.

[0390] "Recording" is the action of saving the match results and related information in a database.

[0391] "Public" means providing match results for access by users and developers via a web interface or the like.

[0392] The "emotion engine" is a system that analyzes the user's facial expressions, tone of voice, input data, etc. to recognize their emotional state.

[0393] "Dynamic interface adjustment" refers to changing the appearance of a web interface and the timing of information presentation according to the user's emotional state.

[0394] This system provides an environment in which generative AI models can tackle problems and compete with each other to find the best solutions, and also combines an emotion engine that recognizes the user's emotions. The system mainly consists of a server, a terminal (for developers), and a user, and each component works in cooperation with each other.

[0395] server

[0396] Registering an AI model

[0397] The server receives the AI ​​model uploaded from the developer's device. After receiving it, the server assigns a unique ID to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. Databases used include MySQL and PostgreSQL. Security checks are also performed to ensure that the model does not contain malicious programs. The security checks use defined rules and a machine learning-based security engine.

[0398] Match Scheduling

[0399] The server schedules the next match using a Python script based on a pre-defined schedule. It selects which AI models will participate and records the match details (task content, evaluation criteria) in a database. The task content is selected from pre-defined prompts, dataset URLs, etc.

[0400] Match execution

[0401] When the match starts, the server uses an API to notify the AI ​​models of the challenge. This notification includes details such as the prompt and the dataset URL. Each AI model generates an answer to the challenge and sends it to the server. The server scores the received answers based on evaluation criteria. Scoring can be done using rule-based evaluation methods or machine learning models.

[0402] Recording and publishing match results

[0403] The server records the match results (scores, generated answers, and evaluation comments) in a database and publishes them on a web interface accessible to developers and users. This web interface is built using HTML / CSS and JavaScript.

[0404] Emotion Engine

[0405] The server is equipped with an emotion engine that recognizes the user's emotions. This emotion engine analyzes the user's facial expressions, tone of voice, input data, etc. while watching a game or viewing game results to recognize the user's emotional state. Examples of emotion engines used include Face++ and IBM Watson.

[0406] Emotion-based interface adjustment

[0407] The server dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine, for example, changing the interface's color scheme or the timing of information presentation if the user is excited.

[0408] Terminal (for developers)

[0409] Submitting an AI model

[0410] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[0411] Check the match results

[0412] Developers can view their model's scores and detailed feedback through a match results interface, which allows them to improve their AI models for the next match.

[0413] User

[0414] Watching a game

[0415] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and the solutions it generates.

[0416] Analyzing the results

[0417] After a match, users can analyze the results in detail, understanding how each AI model solved the problem and how the score was determined based on the evaluation criteria. Furthermore, based on the emotional data recognized by the emotion engine, analysis and feedback are provided according to the user's emotional state.

[0418] Specific examples

[0419] Example 1: Text generation task

[0420] The server sets the task of "generating a poem." For example, it sets a prompt such as "Generate a poem with an autumn landscape theme." Developer A submits AI model A, and developer B submits AI model B. When the match starts, the server notifies each AI model of the task. Each AI model generates a poem and sends it to the server. The server evaluates the generated poems and scores them based on criteria such as literary value and grammatical accuracy. While watching the match, if the user's emotions are high, the interface changes to a brighter color tone.

[0421] Example 2: Image classification task

[0422] The server sets the task of "classifying cat images." Developer C submits AI model C, and developer D submits AI model D. When the match begins, the server notifies each AI model of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server scores the models based on accuracy rate and classification speed. When checking the match results, if the interface recognizes that the user is confused, it displays support information appropriate to the situation.

[0423] In this way, the AI ​​model battle system not only provides a competitive environment between generated AI models, but also adjusts the interface to take the user's emotions into account, providing a more fulfilling experience.

[0424] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0425] Step 1:

[0426] The server receives the AI ​​model from the developer's device. During reception, communication encryption (e.g., HTTPS) is used to ensure data security. The input data is the AI ​​model file uploaded by the developer, and the output is a notification that reception is complete.

[0427] Step 2:

[0428] The server assigns a unique ID to the received AI model. This ID is used to identify the model within the system. The input data is an AI model file, and the output is an AI model with a unique ID.

[0429] Step 3:

[0430] The server stores the AI ​​model and related information (developer name, model name, version, etc.) in a database. The database used is MySQL, PostgreSQL, etc. The input data is the AI ​​model and its related information, and the output is the confirmation information stored in the database.

[0431] Step 4:

[0432] The server performs a security check on the received AI model to ensure it does not contain malicious code. The security engine performs the check using defined rules or machine learning-based methods. The input data is the AI ​​model file, and the output is the result of the security check.

[0433] Step 5:

[0434] The server schedules upcoming matches based on a pre-defined schedule using a Python script. The input data is the schedule information, and the output is the match schedule information.

[0435] Step 6:

[0436] The server retrieves a list of available AI models from the database and selects an AI model to participate in the next match randomly or according to pre-defined conditions. The input data is the stored AI model information, and the output is a list of selected AI models.

[0437] Step 7:

[0438] The server records the details of the match (task content, evaluation criteria, etc.) in a database. The input data is the setting information about the match, and the output is the detailed match information recorded in the database.

[0439] Step 8:

[0440] The server uses an API to notify each AI model of the challenge at the start of the match. This notification includes a prompt and a dataset URL. The input data are the prompt and dataset URL, and the output is confirmation that the notification was received.

[0441] Step 9:

[0442] Each AI model generates a solution to a problem and sends it to the server. The input data is the problem prompt or dataset, and the output is the generated answer.

[0443] Step 10:

[0444] The server scores the received answers based on the evaluation criteria. Scoring can be done using rule-based evaluation methods or machine learning models. The input data are the generated answers, and the output is the scored results.

[0445] Step 11:

[0446] The server records the match results (e.g., scores, generated answers, and evaluation comments) in a database. The input data are the scoring results, and the output is the match results stored in the database.

[0447] Step 12:

[0448] The server publishes the match results through a web interface built using HTML / CSS and JavaScript. The input data are the match results from the database, and the output is the published match results page.

[0449] Step 13:

[0450] The server collects the user's facial expressions, tone of voice, input data, etc. and temporarily stores them in storage. The input data is the emotional state data from the user, and the output is the data stored in storage.

[0451] Step 14:

[0452] The server uses an emotion engine (e.g., Face++ or IBM Watson) to analyze the collected user data and recognize the emotional state. The input data is the emotional state data, and the output is the recognized emotional state.

[0453] Step 15:

[0454] The server dynamically adjusts the web interface based on the recognized emotional state, for example, changing the interface's color scheme or the timing of information presentation if the user is excited. The input data is the recognized emotional state, and the output is the dynamically adjusted interface.

[0455] Through this processing step, the AI ​​model competition system provides a competitive environment for generated AI models and also adjusts the interface to take user emotions into account, providing a more fulfilling experience.

[0456] (Application example 2)

[0457] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0458] Conventional competition systems using generative AI models lack the ability to adjust the interface to take into account user emotions and real-time feedback, resulting in a limited user experience. Furthermore, there is a lack of a means to visualize the competition between generative AI models and effectively communicate the results to users. Specifically, there is a need for a system that can change the interface based on emotional states, recommend appropriate products, and provide these in an integrated manner.

[0459] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for recognizing user emotions, means for adjusting the interface based on the user's emotional state, and means for having generated AI models compete with each other to select the optimal answer. This makes it possible to adjust the interface according to the user's emotional state and recommend optimal products.

[0460] An "AI model" is a collection of artificial intelligence algorithms designed to solve a specific problem.

[0461] A "means for scheduling a match" is a method for determining the date and time of the next match and planning the match for the AI ​​model based on a pre-set schedule.

[0462] The "means of notifying the task" refers to the method of transmitting detailed information about the task to the AI ​​model.

[0463] The "means of receiving the answer" is the mechanism by which the AI ​​model receives the answer generated based on the task.

[0464] A "criteria-based scoring means" is a method of evaluating generated answers according to predetermined criteria and assigning points.

[0465] "Means for recording and publishing match results" means a method for storing match results and publishing them in a form accessible to interested parties.

[0466] "Means for recognizing user emotions" refers to a method for identifying the emotions a user is feeling by analyzing the user's facial expressions, tone of voice, input data, etc.

[0467] "Means for adjusting the interface" refers to a method for dynamically changing the color tone and display content of the screen based on the user's emotional state.

[0468] "Method of having generative AI models compete with each other to select the optimal answer" is a method in which multiple generative AI models tackle the same task and select the best answer from among them.

[0469] This invention is a system that provides an environment in which generative AI models tackle problems and compete with each other for their solutions, and also combines it with an emotion engine that recognizes the user's emotions. This system is mainly composed of a server, a terminal (for developers), and a user, and each component works in cooperation with each other.

[0470] server

[0471] 1. Registering an AI model

[0472] The server receives the AI ​​model uploaded by the developer from the device. After receiving it, it assigns a unique ID and stores the model and related information (developer name, model name, version, etc.) in a database. It also performs a security check to ensure that the model does not contain any malicious programs.

[0473] 2. Match Scheduling

[0474] The server schedules matches based on a pre-set schedule, selects which AI models will participate, and records the tasks and evaluation criteria in a database.

[0475] 3. Match execution

[0476] When the match starts, the server notifies the AI ​​models of the challenge. This notification includes details such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on a set of criteria.

[0477] 4. Recording and publishing match results

[0478] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[0479] 5. Emotion Engine

[0480] The server is equipped with an emotion engine that recognizes the user's emotions. This emotion engine analyzes the user's facial expressions, tone of voice, and input data while watching a game or viewing game results, and recognizes the user's emotional state.

[0481] 6. Emotion-Based Interface Adjustment

[0482] The server dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine, for example, adjusting the interface's color scheme or the timing of information presentation if the user is excited.

[0483] Terminal (for developers)

[0484] 1. Submitting an AI model

[0485] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[0486] 2. Check the match results

[0487] Developers can view their model's scores and detailed feedback through a match result interface, which allows them to improve their AI model for the next match.

[0488] User

[0489] 1. Watching a game

[0490] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[0491] 2. Analysis of results

[0492] After a match, users can analyze the results in detail, understanding how each AI model solved the problem and how the score was determined based on the evaluation criteria. Furthermore, analysis and feedback based on the user's emotional state are provided based on emotional data recognized by the emotion engine.

[0493] Specific examples

[0494] Example 1: Emotion-based product recommendations

[0495] When a user is using the virtual store, their facial expressions are detected in real time via a camera. If the user is determined to be happy, the interface is adjusted to a brighter color tone and products with many positive reviews are recommended. At the same time, Generative AI Model A (specializing in fashion) and Generative AI Model B (specializing in gadgets) compete to recommend the best products for the user.

[0496] Example

[0497] User information: User's age, gender, purchase history

[0498] Camera frame: User's face image

[0499] Prompt statement:

[0500] User Information

[0501] user_profile = {

[0502] 'age': 30,

[0503] 'gender': 'male',

[0504] 'purchase_history': ['laptop', 'smartphone']

[0505] }

[0506] Frames from the camera

[0507] video_frame = cv2.imread('user_face.jpg')

[0508] recommendations, theme = main(user_profile, video_frame)

[0509] print(f'Recommendations: {recommendations}')

[0510] print(f'Interface Theme: {theme}')

[0511] By utilizing this invention, it becomes possible to flexibly adjust the interface according to the user's emotional state and to achieve highly accurate product recommendations through competition between generative AI models.

[0512] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0513] Step 1:

[0514] The server receives AI models uploaded from the device. The input is the AI ​​model file, and the output is a database that stores the model and related information (developer name, model name, version, etc.). Specifically, the server assigns a unique ID to the received AI model and performs security checks to ensure that it does not contain any malicious code.

[0515] Step 2:

[0516] The server schedules matches based on a pre-set schedule. The input is schedule information and a list of registered AI models, and the output is the date and time of the next match, the participating AI models, and a record of the task content. The server selects which AI models will participate and records the task content and evaluation criteria in a database.

[0517] Step 3:

[0518] When the match starts, the server notifies the AI ​​models of the set tasks. The input is detailed task information (such as a prompt or the URL of the dataset), and the output is a task notification to each participating AI model. Specifically, the server sends a notification containing task details to each AI model.

[0519] Step 4:

[0520] Each AI model generates an answer to the task and sends it to the server. The input is detailed information about the task, and the output is the generated answer. The AI ​​model analyzes the received task prompt and processes it to generate an answer.

[0521] Step 5:

[0522] The server scores the received answers based on the evaluation criteria. The input is the answer from each AI model and the pre-defined evaluation criteria, and the output is the score of the answer. The server analyzes the answers according to the evaluation criteria and assigns a score to each answer.

[0523] Step 6:

[0524] The server records the match results in a database and publishes them on a web interface. The input is the match score and answer data, and the output is the published match results. The server stores the match results, scores of each AI model, generated answers, evaluation comments, etc. in a database and makes them accessible to users and developers.

[0525] Step 7:

[0526] The server uses an emotion engine to recognize the user's emotions. The input is the user's facial expression, tone of voice, and input data, and the output is the user's emotional state. Specifically, the server analyzes data from the camera and microphone using the emotion engine to identify the user's emotions.

[0527] Step 8:

[0528] The server adjusts the interface based on the user's emotional state recognized by the emotion engine. The input is the recognized emotional state, and the output is an adjusted interface. For example, if the user is happy, the color tone of the interface or the timing of information presentation can be changed.

[0529] Step 9:

[0530] The terminal (for developers) submits its own AI model to the server. The input is the AI ​​model file created by the developer, and the output is the AI ​​model registered on the server. The terminal uploads the AI ​​model and enters related information through a web interface.

[0531] Step 10:

[0532] Users can check the match results and perform detailed analysis. The input is the match result data published on the server, and the output is the information analyzed by the user. Specifically, users can view the match results and evaluation comments using a web interface.

[0533] Step 11:

[0534] The server selects the optimal answer by having the generative AI models compete with each other. The input is the answers from multiple generative AI models, and the output is the selected optimal answer. The server compares the answers from multiple generative AI models and presents the answer with the highest evaluation to the user.

[0535] This trend will enable the competitive environment between generative AI models and interface adjustments based on user emotions.

[0536] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0537] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0538] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0539] [Second embodiment]

[0540] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0541] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0542] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0543] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0544] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0545] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0546] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0547] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0548] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0549] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0550] In the smart glasses 214, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0551] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0552] This system provides an environment in which generative AIs tackle problems and compete with each other to find the best solutions. The system mainly consists of a server, a terminal (for developers), and a user. The specific functions and operations of each component are explained below.

[0553] server

[0554] 1. Registering an AI model

[0555] The server receives the AI ​​model uploaded from the developer's device. After receiving it, it assigns a unique ID to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. It also performs a security check to ensure that the model does not contain any malicious programs.

[0556] 2. Match Scheduling

[0557] The server schedules the next match based on a pre-defined schedule, selects which AI models will participate, and records match details (e.g., task content, evaluation criteria) in a database.

[0558] 3. Match execution

[0559] When the match starts, the server notifies the AI ​​models of the challenge. The challenge notification includes details such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on the evaluation criteria.

[0560] 4. Recording and publishing match results

[0561] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[0562] Terminal (for developers)

[0563] 1. Submitting an AI model

[0564] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[0565] 2. Check the match results

[0566] Developers can view their model's scores and detailed feedback through a match results interface, which allows them to improve their AI model for the next match.

[0567] User

[0568] 1. Watching a game

[0569] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[0570] 2. Analysis of results

[0571] After the match, users can analyze the results in detail, understanding how each AI model solved the task and how the score based on the evaluation criteria was determined.

[0572] Specific examples

[0573] Example 1: Text generation task

[0574] The server sets the task of "generating a poem." Developer A submits AI model A, and developer B submits AI model B. When the match begins, the server displays a prompt to "generate a poem with an autumn landscape theme." Each AI model generates a poem and submits it to the server. The server evaluates the generated poems and assigns them scores based on criteria such as literary merit and grammatical accuracy.

[0575] Example 2: Image classification task

[0576] The server sets the task of "classifying cat images." Developer C submits AI model C, and developer D submits AI model D. When the match begins, the server notifies the AI ​​model of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server scores the AI ​​models based on accuracy rate and classification speed.

[0577] In this way, the AI ​​Battlefield system provides an environment in which developers can compete against other AI models and a means for users to observe the progress of AI technology in real time.

[0578] The processing flow will be explained below.

[0579] Step 1:

[0580] Uploading an AI model

[0581] Device: Developers upload the AI ​​model from their own device via the web interface, select the model from the file selection screen, and press the send button.

[0582] Step 2:

[0583] Security checks for AI models

[0584] Server: Receives the uploaded AI model and performs security checks to ensure it does not contain malicious code or viruses.

[0585] Step 3:

[0586] Registering an AI model

[0587] Server: After security checks are complete, it assigns a unique ID to the AI ​​model and stores the model and its metadata (developer name, model name, version, etc.) in a database.

[0588] Step 4:

[0589] Match Scheduling

[0590] Server: Schedules the next match based on a pre-set schedule. At this time, it selects the participating AI models and records the match details (task content, evaluation criteria) in a database.

[0591] Step 5:

[0592] Assignment notifications

[0593] Server: When the match start time arrives, the server notifies the participating AI models of the challenge, including details such as the prompt and the dataset URL.

[0594] Step 6:

[0595] Generate answers

[0596] Terminal: Each participating AI model generates an answer to the task and sends the result to the server. The answer generation must be completed within a specified time.

[0597] Step 7:

[0598] Receiving answers

[0599] Server: Receives the answers sent from each AI model and stores them in a database. Monitors the reception status and confirms that all answers have been received.

[0600] Step 8:

[0601] Scoring answers

[0602] Server: Scores the received answers based on evaluation criteria, such as originality, accuracy, and efficiency. Stores the scoring results in a database.

[0603] Step 9:

[0604] Recording match results

[0605] Server: Generates match results based on the scoring results and records them in a database, including the scores of each AI model, generated answers, and evaluation comments.

[0606] Step 10:

[0607] Publication of match results

[0608] Server: The server publishes the match results on a web interface accessible to developers and users. The results page displays the scores and evaluation comments for each AI model.

[0609] Step 11:

[0610] Check the match results

[0611] Device: Developers can view match results, scores, and detailed feedback from their own devices through a web interface, which can be used to improve the game for the next time.

[0612] Step 12:

[0613] Watching a game

[0614] Users: Users can watch the matches in real time and observe the performance of each AI model, and after the match, they can also perform detailed analysis of the results.

[0615] Example 1

[0616] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0617] Conventional competition systems using AI models lacked an efficient workflow, from model registration to running matches, publishing results, and analyzing them. Furthermore, they often lacked important features such as security checks and detailed feedback, resulting in issues of low usability and reliability for AI developers. There was a need for an effective system that could solve these issues, promote competition between AI models, and present technological progress to users in real time.

[0618] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0619] In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for performing security checks, means for assigning unique identifiers to AI models, means for users to watch matches in real time through a web interface, means for analyzing match results in detail, means for providing detailed feedback on generated answers, and means for providing developers with information to improve their AI models. This realizes an environment in which a series of processes, from registering AI models to detailed analysis of match results, can be executed efficiently and reliably, and makes it possible to provide AI developers with useful information.

[0620] An "AI model" is software designed to perform a specific task using artificial intelligence techniques.

[0621] A "server" refers to a computer system that provides services to other devices over a network.

[0622] "Device" means an electronic device used by a user or developer to access the Service.

[0623] "User" means any person or organization that uses the system or service.

[0624] A "challenge" is a task or problem that an AI model is given to solve.

[0625] An "answer" is the result or response that an AI model generates based on a task.

[0626] "Evaluation criteria" refers to the criteria or scale used to score the generated answers.

[0627] "Scoring" refers to assigning points to generated answers based on evaluation criteria.

[0628] A "database" refers to a system for efficiently storing, managing, and retrieving information.

[0629] A "security check" is an inspection to ensure that a system or software does not contain malicious programs.

[0630] A "unique identifier" is a number or code assigned to an element to uniquely identify it from all other elements.

[0631] "Web interface" refers to a web page or application that can be accessed through a browser.

[0632] "Feedback" refers to evaluations and comments on the AI ​​model's answers, which can be used to improve the model.

[0633] "Developers" refers to the engineers and programmers who create AI models and submit them to the system.

[0634] "Real-time" means that an event is processed and displayed immediately, with almost no delay, after it occurs.

[0635] "Registration" refers to the act of adding new information or models to the system.

[0636] "Scheduling" refers to setting a timetable for the execution of a particular event or task.

[0637] This invention is a system that provides an environment in which generative AI models tackle problems and compete with each other to find the best solutions. This system is mainly composed of a server, a terminal (for developers), and a user.

[0638] server

[0639] The server plays a central role in managing and operating the entire system using multiple pieces of hardware and software. The main functions of the server are as follows:

[0640] Registering an AI model

[0641] The server receives the AI ​​model uploaded from the developer's device and assigns it a unique identifier. The received AI model and related information (developer name, model name, version, etc.) are stored in a database. A security check is performed to ensure that the model does not contain any malicious programs.

[0642] Match Scheduling

[0643] The server schedules the next match based on a pre-defined schedule, selects the participating AI models, and records match details (e.g., task content, evaluation criteria) in a database.

[0644] Match execution

[0645] When the match starts, the server notifies the AI ​​models of the challenge. This challenge notification includes detailed information such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on the evaluation criteria.

[0646] Recording and publishing match results

[0647] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[0648] As a concrete example, the server can set a challenge to "generate a poem." For example, developer A submits AI model A, and developer B submits AI model B. When the match starts, the server notifies the participants with a prompt to "generate a poem on the theme of autumn scenery." Each AI model generates a poem and submits it to the server. The server evaluates the generated poems and assigns them scores based on criteria such as literary merit and grammatical accuracy.

[0649] Terminal (for developers)

[0650] The terminals are used by developers to submit AI models and check match results. Developers can use their own terminals to submit AI models to the server through a web interface and check match results.

[0651] Submitting an AI model

[0652] Developers can submit AI models to the server via a web interface using their own devices, and can enter model summary and version information when submitting.

[0653] Check the match results

[0654] Developers can view match results through a web interface, view their model's scores and detailed feedback, and improve their AI models for the next match.

[0655] User

[0656] Users use the system primarily to watch games and analyze results.

[0657] Watching a game

[0658] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[0659] Analyzing the results

[0660] After the match, users can analyze the results in detail, understanding how each AI model solved the task and how the score based on the evaluation criteria was determined.

[0661] As a concrete example, if the server sets the task of "classifying images of cats," developer C submits AI model C, and developer D submits AI model D. When the time for the match to start arrives, the server notifies it of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server then scores them based on accuracy rate and classification speed.

[0662] Examples of prompts include "Generate a poem on the theme of autumn scenery" and "Determine whether the image below is a cat."

[0663] The system promotes competition between AI models and provides an effective environment where developers and users can obtain information tailored to their respective goals.

[0664] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0665] Step 1: Registering an AI model

[0666] The server receives the AI ​​model from the developer's device. Specifically, the developer uploads the AI ​​model file through a web interface, and the server receives it. The input data is the AI ​​model file, and the output is the model's unique identifier and registration results. After receiving it, the server assigns a unique identifier to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. In addition, it performs security checks to ensure the AI ​​model does not contain malicious programs. Security checks include virus scans and static code analysis.

[0667] Step 2: Scheduling matches

[0668] The server schedules the next match based on a pre-set schedule. The input data is a list of AI models in the database and schedule information, and the output is detailed match information. Specifically, the server selects models to participate in the match from the list of AI models stored in the database. Past performance and random selection are used as criteria. Match details include task content and evaluation criteria, and this information is recorded in the database.

[0669] Step 3: Assignment notification

[0670] When the match starts, the server notifies the AI ​​models of the task. The input data is the task content and a list of participating AI models, and the output is the notification result. The server notifies each AI model of detailed information such as the prompt text and the URL of the dataset. Specifically, the server communicates the task to each model using an API call.

[0671] Step 4: Generate and submit answers

[0672] Each AI model generates an answer to the task and sends it to the server. The input data is the task notification content, and the output is the generated answer. Specifically, the AI ​​model generates an answer based on the received prompt. For example, it generates a poem in response to the prompt "Generate a poem on the theme of autumn scenery." The generated answer is then sent to the server.

[0673] Step 5: Receiving and evaluating answers

[0674] The server receives the answers from the AI ​​model and scores them based on evaluation criteria. The input data is the received answers, and the output is the evaluation results and score. The server scores the received answers based on evaluation criteria such as literary value and grammatical accuracy.

[0675] Step 6: Record and publish match results

[0676] The server records the match results in a database and makes them publicly available through a web interface. The input data are the evaluation results and scores, and the output is the publicly available information. The server stores the scores, generated answers, and evaluation comments of each AI model in a database and makes them publicly available through a web interface accessible to developers and users.

[0677] Step 7: Submit your AI model

[0678] Developers submit AI models through a web interface using their own devices. The input data is the AI ​​model file, its summary, and version information, and the output is the submission results. Developers select and upload the model file, enter the model summary and version information, and submit.

[0679] Step 8: Check the match results

[0680] Developers check the match results through a web interface. The input data is the match result information, and the output is the displayed score and feedback. Developers can check the scores and detailed feedback of their AI models and use them to improve their AI models for the next match.

[0681] Step 9: Watch the game

[0682] Users use an interface that allows them to watch the game in real time. The input data is game information updated in real time, and the output is the viewing experience. Users can observe live how each AI model tackles the problem and what solutions it generates.

[0683] Step 10: Analyze the results

[0684] Users can analyze the match results in detail after the match is over. The input data is the published match result information, and the output is the analysis results. Users can analyze the answers generated by each AI model and their scores based on the evaluation criteria to understand how they solved the problem. By referring to the evaluation comments, users can grasp the strengths and areas for improvement of the AI ​​model.

[0685] (Application example 1)

[0686] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0687] Previous systems utilizing generative AI models simply evaluated and compared the performance of AI models, lacking the ability to incorporate user participation and evaluation in real time. Furthermore, there were insufficient means for developers to receive detailed feedback, making it time-consuming and labor-intensive to improve AI models. Furthermore, there was a lack of platforms where users could evaluate and enjoy generated content, limiting opportunities for the general public to understand and enjoy technological advances in generative AI.

[0688] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0689] In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for users to view and evaluate generated content in real time, and means for developers to receive detailed feedback on the models. This allows the technology of generative AI models to be utilized while incorporating users' real-time evaluations, allowing developers to quickly improve their models based on detailed feedback. Furthermore, the server functions as a platform where general users can enjoy content created by generative AI, thereby promoting widespread understanding of advances in generative AI technology.

[0690] An "AI model" is an algorithm and database that uses artificial intelligence technology to generate answers and predictions for specific problems.

[0691] A "problem" is a specific problem or task that an AI model must address, and the resulting solutions are evaluated against it.

[0692] "Answer" refers to the output generated by an AI model based on a task.

[0693] "Users" are participants who use the system to view and evaluate generated content in real time.

[0694] "Developer" refers to a person or organization that develops an AI model and registers it in the system to test its performance.

[0695] "Means for notifying tasks" refers to a function that allows the server to communicate the tasks that need to be solved to the AI ​​model.

[0696] "Means for receiving answers" refers to the function for receiving answers generated from the AI ​​model on the server side.

[0697] "Means for scoring" refers to a function that assigns points to received answers based on evaluation criteria.

[0698] "Means for recording and publishing match results" refers to the ability to store the evaluation results of scored answers and display them in a form accessible to users and developers.

[0699] The "means for viewing and rating in real time" is a function that allows users to view generated content in real time and rate it on the spot.

[0700] "Means for receiving feedback" is a function that allows developers to receive detailed comments and suggestions for improvement regarding the AI ​​model's match results and user evaluations.

[0701] This invention is a system for pitting generative AI models against each other in specific challenges, allowing users to view and evaluate the generated content. Specifically, it consists of a server, a developer terminal, and a user interface.

[0702] server

[0703] 1. Registering the AI ​​model:

[0704] The server receives the AI ​​model provided by the developer, assigns a unique ID, and stores it in a database. It also performs security checks to ensure that no malicious programs are included. This process can be implemented using Python and Flask, and TensorFlow and SQLite are used to store the AI ​​model.

[0705] 2. Match Scheduling:

[0706] The server schedules matches, determines the next match time, and selects the AI ​​models that will participate. This information is recorded in a database and notifications are sent as needed. The scheduling logic is automatic, based on pre-defined times and conditions.

[0707] 3. Assignment Notification:

[0708] When the game starts, the AI ​​model is notified of the task. The task includes a prompt and the URL of the required dataset. For example, the prompt might be "Generate a poem based on an autumn landscape." This process is notified to the AI ​​model via an HTTP request.

[0709] 4. Receiving and Scoring Answers:

[0710] The answers generated by each AI model are sent to the server, where they are then scored based on evaluation criteria. Scoring evaluates the quality of the generated content and its suitability for the task. This process uses natural language processing (NLP) technology and evaluation algorithms.

[0711] 5. Recording and Publication of Match Results:

[0712] Match results are recorded in a database and made publicly accessible to users and developers, allowing users to rate the generated content and developers to receive feedback.

[0713] Developer Device

[0714] 1. Submitting an AI model:

[0715] Developers submit AI models to the server from their own devices, providing details such as the model name, version information, and developer details. Submissions are done through a web-based interface.

[0716] 2. Match result confirmation and feedback:

[0717] Developers have an interface that allows them to view match results and detailed evaluation feedback for each model, providing them with immediate information they need to improve their models.

[0718] User

[0719] 1. Real-time viewing and evaluation:

[0720] Users can watch the progress of the match in real time. The generated content is displayed in real time, and users can rate and comment on it on the spot. This function enables user-participation in evaluation.

[0721] 2. Analysis of match results:

[0722] After the match, users can analyze the evaluation results and scoring details of the generated content, helping them see how their own evaluations are reflected and understand the technological progress of the generating AI.

[0723] Specific examples

[0724] One weekend, the task is to "generate a poem with the theme of autumn scenery." Developer A's model generates a poem that "speaks of autumn, with red and yellow leaves fluttering in the air, and sunsets that color the sky. A poem that resonates with the heart, like a whisper of the wind." Meanwhile, Developer B's model generates a poem that "speaks of autumn scenery that evokes warm memories in the heart, with cold winds blowing." Each poem is evaluated by users and assigned a score.

[0725] Prompt Sentence Examples

[0726] "Generate a poem with the theme of autumn scenery. We're looking for poems that incorporate the beautiful colors, the sound of the wind, the glow of the sunset, etc."

[0727] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0728] Step 1: Registering an AI model

[0729] Input: Developers upload the AI ​​model file from their own device to the server.

[0730] Processing: The server receives the uploaded AI model, assigns a unique ID, and records related information such as the model name, developer name, and version in a database. It also performs security checks to ensure there are no malicious programs.

[0731] Output: The AI ​​model is saved in a database, along with the model's unique ID and related information.

[0732] Step 2: Scheduling the match

[0733] Input: The server plans the next match based on a predefined match schedule.

[0734] Processing: The server selects participating AI models according to the schedule and records the match date, time, task content, and evaluation criteria in a database.

[0735] Output: The next match information is saved in the database and ready to be notified.

[0736] Step 3: Issue notification

[0737] Input: As the start time of the match approaches, the server notifies the AI ​​model of the task.

[0738] Processing: The server sends each AI model a prompt and the URL of the dataset via an HTTP request. For example, the prompt might be "Generate a poem based on an autumn landscape."

[0739] Output: Each AI model receives the task information.

[0740] Step 4: Generate and submit your answers

[0741] Input: The AI ​​model generates an answer based on the received challenge.

[0742] Processing: The AI ​​model follows the prompt, uses natural language processing techniques to generate poetry or other content, and sends the answer to the server.

[0743] Output: The generated answer is sent to the server.

[0744] Step 5: Receiving and scoring answers

[0745] Input: The server receives the answers sent by each AI model.

[0746] Processing: The server collects the answers and scores them based on a set of criteria, including literary merit and grammatical accuracy.

[0747] Output: The answer and score of each AI model are recorded in a database.

[0748] Step 6: Record and publish match results

[0749] Input: After scoring is complete, the server compiles the match results.

[0750] Processing: The server records the match results in a database and exposes them through an interface, making this information accessible to users and developers.

[0751] Output: The match results are published and available for users and developers to view.

[0752] Step 7: Real-time viewing and evaluation by users

[0753] Input: Users access the generated content to watch it in real time.

[0754] Processing: The server displays the generated content to the user in real time and provides a mechanism for the user to immediately submit ratings and comments.

[0755] Output: User ratings and comments are recorded in the system.

[0756] Step 8: Detailed analysis of match results

[0757] Input: After the match is over, users and developers have access to analyze the match results.

[0758] Processing: The server displays detailed match results and feedback, and provides tools that allow users to analyze the performance of each AI model.

[0759] Output: Detailed match results and analytical information are displayed to users and developers.

[0760] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0761] This system provides an environment in which generative AIs can tackle problems and compete with each other to find the best answers, and also combines an emotion engine that recognizes the user's emotions. The system mainly consists of a server, a terminal (for developers), and a user, and each component works in cooperation. The specific functions and operations of each component are explained below.

[0762] server

[0763] 1. Registering an AI model

[0764] The server receives the AI ​​model uploaded from the developer's device. After receiving it, the server assigns a unique ID to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. It also performs security checks to ensure that the model does not contain any malicious programs.

[0765] 2. Match Scheduling

[0766] The server schedules the next match based on a pre-defined schedule, selects which AI models will participate, and records match details (task content, evaluation criteria) in a database.

[0767] 3. Match execution

[0768] When the match starts, the server notifies the AI ​​models of the challenge. This notification includes details such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on the evaluation criteria.

[0769] 4. Recording and publishing match results

[0770] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[0771] 5. Emotion Engine

[0772] The server is equipped with an emotion engine that recognizes the user's emotions. This emotion engine analyzes the user's facial expressions, tone of voice, input data, etc. while watching a game or viewing game results, and recognizes the user's emotional state.

[0773] 6. Emotion-Based Interface Adjustment

[0774] The server dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine, for example, adjusting the interface's color scheme and the timing of information presentation if the user is excited.

[0775] Terminal (for developers)

[0776] 1. Submitting an AI model

[0777] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[0778] 2. Check the match results

[0779] Developers can view their model's scores and detailed feedback through a match results interface, which allows them to improve their AI model for the next match.

[0780] User

[0781] 1. Watching a game

[0782] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[0783] 2. Analysis of results

[0784] After a match, users can analyze the results in detail, understanding how each AI model solved the problem and how the score was determined based on the evaluation criteria. Furthermore, based on the emotional data recognized by the emotion engine, analysis and feedback are provided according to the user's emotional state.

[0785] Specific examples

[0786] Example 1: Text generation task

[0787] The server sets the task of "generating a poem." Developer A submits AI model A, and developer B submits AI model B. When the match begins, the server displays a prompt to "generate a poem with an autumn landscape theme." Each AI model generates a poem and sends it to the server. The server evaluates the generated poems and assigns them scores based on criteria such as literary value and grammatical accuracy. While watching the match, if the user's emotions are high, the interface changes to a brighter color tone.

[0788] Example 2: Image classification task

[0789] The server sets the task of "classifying cat images." Developer C submits AI model C, and developer D submits AI model D. When the match begins, the server notifies the user of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server scores the users based on accuracy rate and classification speed. When checking the match results, if the user is detected as confused, the interface displays additional support information appropriate to the situation.

[0790] In this way, the AI ​​Battlefield system not only provides a competitive environment between generative AI models, but also adjusts the interface to take user emotions into account, providing a more fulfilling experience.

[0791] The processing flow will be explained below.

[0792] MODE FOR CARRYING OUT THE INVENTION

[0793] Server Operation

[0794] Step 1:

[0795] Receiving AI model uploads

[0796] Server: Receives the AI ​​model uploaded from the developer's device. After receiving it, it starts a security check.

[0797] Step 2:

[0798] Security Check

[0799] Server: Performs security checks on the received AI model to check for malicious code and viruses.

[0800] Step 3:

[0801] Registering an AI model

[0802] Server: Once the security check is complete, it assigns a unique ID to the AI ​​model and stores the model and its metadata (developer name, model name, version) in a database.

[0803] Step 4:

[0804] Match Scheduling

[0805] Server: Schedules the next match based on a pre-set schedule, selects participating AI models, and records detailed match information (task content, evaluation criteria) in a database.

[0806] Step 5:

[0807] Assignment notifications

[0808] Server: When the match start time arrives, the server notifies the participating AI models of the challenge, including details such as the prompt and the dataset URL.

[0809] Step 6:

[0810] Receiving answers

[0811] Server: Receives answers from each participating AI model and stores them in a database. Monitors the reception status and confirms that all answers have been received.

[0812] Step 7:

[0813] Scoring

[0814] Server: Scores the received answers based on criteria such as originality, accuracy, and efficiency.

[0815] Step 8:

[0816] Recording the results

[0817] Server: Generates match results based on the scoring results and records them in a database, including the scores of each AI model, generated answers, and evaluation comments.

[0818] Step 9:

[0819] Publication of match results

[0820] Server: The server publishes the match results on a web interface accessible to developers and users. The results page displays the scores and evaluation comments for each AI model.

[0821] Emotion Engine Operation

[0822] Step 10:

[0823] User Emotion Recognition

[0824] Server: Uses an emotion engine to recognize emotions from the user's facial expressions, tone of voice, and input data.

[0825] Step 11:

[0826] Interface Adjustments

[0827] Server: Dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine. For example, if the user is excited, the color scheme of the interface or the timing of information presentation is adjusted.

[0828] User Actions

[0829] Step 12:

[0830] Watching a game

[0831] Users: Watch the match in real time and observe the performance of each AI model.

[0832] Step 13:

[0833] Analyzing the results

[0834] User: After the match, the results are analyzed in detail. Based on the emotional data recognized by the emotion engine, analysis and feedback are provided according to the user's emotional state.

[0835] Device (developer) operation

[0836] Step 14:

[0837] Submitting an AI model

[0838] Device: Developers submit AI models from their own devices through a web interface, entering model overview and version information.

[0839] Step 15:

[0840] Check the match results

[0841] On-device: Developers can view their model's scores and detailed feedback through an interface that allows them to check the match results and make adjustments to improve their model for next time.

[0842] Specific examples

[0843] Example 1: Text generation task

[0844] Step 1:

[0845] Assignment settings

[0846] Server: The server sets the task of "creating a poem."

[0847] Step 2:

[0848] Submitting an AI model

[0849] Terminal: Developer A submits AI model A, and developer B submits AI model B.

[0850] Step 3:

[0851] Assignment notifications

[0852] Server: When the match start time arrives, the server will announce the prompt to "Generate a poem with an autumn landscape theme."

[0853] Step 4:

[0854] Generating and receiving answers

[0855] Terminal: Each AI model generates a poem and sends it to the server, which receives the answer and evaluates it.

[0856] Step 5:

[0857] Scoring

[0858] Server: Evaluates the generated poems, scoring them on criteria such as literary merit and grammatical accuracy.

[0859] Step 6:

[0860] emotion recognition

[0861] Server: While watching the game, the emotion engine recognizes the user's emotions.

[0862] Step 7:

[0863] Interface Adjustments

[0864] Server: If the user is emotionally excited, change the interface color to a brighter tone.

[0865] Example 2: Image classification task

[0866] Step 1:

[0867] Assignment settings

[0868] Server: The server sets the task "classify images of cats."

[0869] Step 2:

[0870] Submitting an AI model

[0871] Terminal: Developer C submits AI model C, and developer D submits AI model D.

[0872] Step 3:

[0873] Assignment notifications

[0874] Server: When the match starts, a series of cat images will be displayed.

[0875] Step 4:

[0876] Generating and receiving answers

[0877] Terminal: Each AI model classifies an image and sends the results to the server, which receives and evaluates the answers.

[0878] Step 5:

[0879] Scoring

[0880] Server: Scoring is based on accuracy rate and classification speed.

[0881] Step 6:

[0882] emotion recognition

[0883] Server: When checking the match results, the emotion engine recognizes the user's emotions.

[0884] Step 7:

[0885] Interface Adjustments

[0886] Server: If the user is detected as confused, the interface displays additional support information.

[0887] In this way, the AI ​​Battlefield system provides a competitive environment for generative AI models and adjusts the interface to take user emotions into account, providing a more fulfilling experience.

[0888] Example 2

[0889] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0890] Conventional AI model competition systems simply record and publish the results of AI model competitions, without adjusting the interface to take into account the user's emotional state. This resulted in a lack of improvement in the user experience and reduced engagement with the system. Furthermore, security checks and scheduling of AI models were not performed efficiently, leaving the system lacking reliability and operational efficiency.

[0891] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for analyzing the user's emotional state, and means for dynamically adjusting the interface based on the user's emotional state. This makes it possible to provide the results of AI model matches in a real-time and reliable manner and to improve the user experience.

[0892] An "AI model" is an artificial intelligence software system that has been trained to perform a specific task.

[0893] A "match" is an event in which multiple AI models compete against each other based on a given task and their performance is evaluated.

[0894] A "problem" is a specific problem or requirement that the AI ​​model must solve, such as text generation or image classification.

[0895] "Notification" is an action in which the server communicates details of the problem to the AI ​​model.

[0896] An "answer" is the resulting data generated by an AI model based on a task.

[0897] "Evaluation criteria" are indicators and rules for evaluating answers generated by an AI model.

[0898] "Scoring" is the process of numerically evaluating an AI model's answer based on evaluation criteria.

[0899] "Match results" are data that include the evaluation results of each AI model's answers.

[0900] "Recording" is the action of saving the match results and related information in a database.

[0901] "Public" means providing match results for access by users and developers via a web interface or the like.

[0902] The "emotion engine" is a system that analyzes the user's facial expressions, tone of voice, input data, etc. to recognize their emotional state.

[0903] "Dynamic interface adjustment" refers to changing the appearance of a web interface and the timing of information presentation according to the user's emotional state.

[0904] This system provides an environment in which generative AI models can tackle problems and compete with each other to find the best solutions, and also combines an emotion engine that recognizes the user's emotions. The system mainly consists of a server, a terminal (for developers), and a user, and each component works in cooperation with each other.

[0905] server

[0906] Registering an AI model

[0907] The server receives the AI ​​model uploaded from the developer's device. After receiving it, the server assigns a unique ID to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. Databases used include MySQL and PostgreSQL. Security checks are also performed to ensure that the model does not contain malicious programs. The security checks use defined rules and a machine learning-based security engine.

[0908] Match Scheduling

[0909] The server schedules the next match using a Python script based on a pre-defined schedule. It selects which AI models will participate and records the match details (task content, evaluation criteria) in a database. The task content is selected from pre-defined prompts, dataset URLs, etc.

[0910] Match execution

[0911] When the match starts, the server uses an API to notify the AI ​​models of the challenge. This notification includes details such as the prompt and the dataset URL. Each AI model generates an answer to the challenge and sends it to the server. The server scores the received answers based on evaluation criteria. Scoring can be done using rule-based evaluation methods or machine learning models.

[0912] Recording and publishing match results

[0913] The server records the match results (scores, generated answers, and evaluation comments) in a database and publishes them on a web interface accessible to developers and users. This web interface is built using HTML / CSS and JavaScript.

[0914] Emotion Engine

[0915] The server is equipped with an emotion engine that recognizes the user's emotions. This emotion engine analyzes the user's facial expressions, tone of voice, input data, etc. while watching a game or viewing game results to recognize the user's emotional state. Examples of emotion engines used include Face++ and IBM Watson.

[0916] Emotion-based interface adjustment

[0917] The server dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine, for example, changing the interface's color scheme or the timing of information presentation if the user is excited.

[0918] Terminal (for developers)

[0919] Submitting an AI model

[0920] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[0921] Check the match results

[0922] Developers can view their model's scores and detailed feedback through a match results interface, which allows them to improve their AI models for the next match.

[0923] User

[0924] Watching a game

[0925] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and the solutions it generates.

[0926] Analyzing the results

[0927] After a match, users can analyze the results in detail, understanding how each AI model solved the problem and how the score was determined based on the evaluation criteria. Furthermore, based on the emotional data recognized by the emotion engine, analysis and feedback are provided according to the user's emotional state.

[0928] Specific examples

[0929] Example 1: Text generation task

[0930] The server sets the task of "generating a poem." For example, it sets a prompt such as "Generate a poem with an autumn landscape theme." Developer A submits AI model A, and developer B submits AI model B. When the match starts, the server notifies each AI model of the task. Each AI model generates a poem and sends it to the server. The server evaluates the generated poems and scores them based on criteria such as literary value and grammatical accuracy. While watching the match, if the user's emotions are high, the interface changes to a brighter color tone.

[0931] Example 2: Image classification task

[0932] The server sets the task of "classifying cat images." Developer C submits AI model C, and developer D submits AI model D. When the match begins, the server notifies each AI model of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server scores the models based on accuracy rate and classification speed. When checking the match results, if the interface recognizes that the user is confused, it displays support information appropriate to the situation.

[0933] In this way, the AI ​​model battle system not only provides a competitive environment between generated AI models, but also adjusts the interface to take the user's emotions into account, providing a more fulfilling experience.

[0934] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0935] Step 1:

[0936] The server receives the AI ​​model from the developer's device. During reception, communication encryption (e.g., HTTPS) is used to ensure data security. The input data is the AI ​​model file uploaded by the developer, and the output is a notification that reception is complete.

[0937] Step 2:

[0938] The server assigns a unique ID to the received AI model. This ID is used to identify the model within the system. The input data is an AI model file, and the output is an AI model with a unique ID.

[0939] Step 3:

[0940] The server stores the AI ​​model and related information (developer name, model name, version, etc.) in a database. The database used is MySQL, PostgreSQL, etc. The input data is the AI ​​model and its related information, and the output is the confirmation information stored in the database.

[0941] Step 4:

[0942] The server performs a security check on the received AI model to ensure it does not contain malicious code. The security engine performs the check using defined rules or machine learning-based methods. The input data is the AI ​​model file, and the output is the result of the security check.

[0943] Step 5:

[0944] The server schedules upcoming matches based on a pre-defined schedule using a Python script. The input data is the schedule information, and the output is the match schedule information.

[0945] Step 6:

[0946] The server retrieves a list of available AI models from the database and selects an AI model to participate in the next match randomly or according to pre-defined conditions. The input data is the stored AI model information, and the output is a list of selected AI models.

[0947] Step 7:

[0948] The server records the details of the match (task content, evaluation criteria, etc.) in a database. The input data is the setting information about the match, and the output is the detailed match information recorded in the database.

[0949] Step 8:

[0950] The server uses an API to notify each AI model of the challenge at the start of the match. This notification includes a prompt and a dataset URL. The input data are the prompt and dataset URL, and the output is confirmation that the notification was received.

[0951] Step 9:

[0952] Each AI model generates a solution to a problem and sends it to the server. The input data is the problem prompt or dataset, and the output is the generated answer.

[0953] Step 10:

[0954] The server scores the received answers based on the evaluation criteria. Scoring can be done using rule-based evaluation methods or machine learning models. The input data are the generated answers, and the output is the scored results.

[0955] Step 11:

[0956] The server records the match results (e.g., scores, generated answers, and evaluation comments) in a database. The input data are the scoring results, and the output is the match results stored in the database.

[0957] Step 12:

[0958] The server publishes the match results through a web interface built using HTML / CSS and JavaScript. The input data are the match results from the database, and the output is the published match results page.

[0959] Step 13:

[0960] The server collects the user's facial expressions, tone of voice, input data, etc. and temporarily stores them in storage. The input data is the emotional state data from the user, and the output is the data stored in storage.

[0961] Step 14:

[0962] The server uses an emotion engine (e.g., Face++ or IBM Watson) to analyze the collected user data and recognize the emotional state. The input data is the emotional state data, and the output is the recognized emotional state.

[0963] Step 15:

[0964] The server dynamically adjusts the web interface based on the recognized emotional state, for example, changing the interface's color scheme or the timing of information presentation if the user is excited. The input data is the recognized emotional state, and the output is the dynamically adjusted interface.

[0965] Through this processing step, the AI ​​model competition system provides a competitive environment for generated AI models and also adjusts the interface to take user emotions into account, providing a more fulfilling experience.

[0966] (Application example 2)

[0967] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0968] Conventional competition systems using generative AI models lack the ability to adjust the interface to take into account user emotions and real-time feedback, resulting in a limited user experience. Furthermore, there is a lack of a means to visualize the competition between generative AI models and effectively communicate the results to users. Specifically, there is a need for a system that can change the interface based on emotional states, recommend appropriate products, and provide these in an integrated manner.

[0969] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for recognizing user emotions, means for adjusting the interface based on the user's emotional state, and means for having generated AI models compete with each other to select the optimal answer. This makes it possible to adjust the interface according to the user's emotional state and recommend optimal products.

[0970] An "AI model" is a collection of artificial intelligence algorithms designed to solve a specific problem.

[0971] A "means for scheduling a match" is a method for determining the date and time of the next match and planning the match for the AI ​​model based on a pre-set schedule.

[0972] The "means of notifying the task" refers to the method of transmitting detailed information about the task to the AI ​​model.

[0973] The "means of receiving the answer" is the mechanism by which the AI ​​model receives the answer generated based on the task.

[0974] A "criteria-based scoring means" is a method of evaluating generated answers according to predetermined criteria and assigning points.

[0975] "Means for recording and publishing match results" means a method for storing match results and publishing them in a form accessible to interested parties.

[0976] "Means for recognizing user emotions" refers to a method for identifying the emotions a user is feeling by analyzing the user's facial expressions, tone of voice, input data, etc.

[0977] "Means for adjusting the interface" refers to a method for dynamically changing the color tone and display content of the screen based on the user's emotional state.

[0978] "Method of having generative AI models compete with each other to select the optimal answer" is a method in which multiple generative AI models tackle the same task and select the best answer from among them.

[0979] This invention is a system that provides an environment in which generative AI models tackle problems and compete with each other for their solutions, and also combines it with an emotion engine that recognizes the user's emotions. This system is mainly composed of a server, a terminal (for developers), and a user, and each component works in cooperation with each other.

[0980] server

[0981] 1. Registering an AI model

[0982] The server receives the AI ​​model uploaded by the developer from the device. After receiving it, it assigns a unique ID and stores the model and related information (developer name, model name, version, etc.) in a database. It also performs a security check to ensure that the model does not contain any malicious programs.

[0983] 2. Match Scheduling

[0984] The server schedules matches based on a pre-set schedule, selects which AI models will participate, and records the tasks and evaluation criteria in a database.

[0985] 3. Match execution

[0986] When the match starts, the server notifies the AI ​​models of the challenge. This notification includes details such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on a set of criteria.

[0987] 4. Recording and publishing match results

[0988] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[0989] 5. Emotion Engine

[0990] The server is equipped with an emotion engine that recognizes the user's emotions. This emotion engine analyzes the user's facial expressions, tone of voice, and input data while watching a game or viewing game results, and recognizes the user's emotional state.

[0991] 6. Emotion-Based Interface Adjustment

[0992] The server dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine, for example, adjusting the interface's color scheme or the timing of information presentation if the user is excited.

[0993] Terminal (for developers)

[0994] 1. Submitting an AI model

[0995] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[0996] 2. Check the match results

[0997] Developers can view their model's scores and detailed feedback through a match result interface, which allows them to improve their AI model for the next match.

[0998] User

[0999] 1. Watching a game

[1000] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[1001] 2. Analysis of results

[1002] After a match, users can analyze the results in detail, understanding how each AI model solved the problem and how the score was determined based on the evaluation criteria. Furthermore, analysis and feedback based on the user's emotional state are provided based on emotional data recognized by the emotion engine.

[1003] Specific examples

[1004] Example 1: Emotion-based product recommendations

[1005] When a user is using the virtual store, their facial expressions are detected in real time via a camera. If the user is determined to be happy, the interface is adjusted to a brighter color tone and products with many positive reviews are recommended. At the same time, Generative AI Model A (specializing in fashion) and Generative AI Model B (specializing in gadgets) compete to recommend the best products for the user.

[1006] Example

[1007] User information: User's age, gender, purchase history

[1008] Camera frame: User's face image

[1009] Prompt statement:

[1010] User Information

[1011] user_profile = {

[1012] 'age': 30,

[1013] 'gender': 'male',

[1014] 'purchase_history': ['laptop', 'smartphone']

[1015] }

[1016] Frames from the camera

[1017] video_frame = cv2.imread('user_face.jpg')

[1018] recommendations, theme = main(user_profile, video_frame)

[1019] print(f'Recommendations: {recommendations}')

[1020] print(f'Interface Theme: {theme}')

[1021] By utilizing this invention, it becomes possible to flexibly adjust the interface according to the user's emotional state and to achieve highly accurate product recommendations through competition between generative AI models.

[1022] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1023] Step 1:

[1024] The server receives AI models uploaded from the device. The input is the AI ​​model file, and the output is a database that stores the model and related information (developer name, model name, version, etc.). Specifically, the server assigns a unique ID to the received AI model and performs security checks to ensure that it does not contain any malicious code.

[1025] Step 2:

[1026] The server schedules matches based on a pre-set schedule. The input is schedule information and a list of registered AI models, and the output is the date and time of the next match, the participating AI models, and a record of the task content. The server selects which AI models will participate and records the task content and evaluation criteria in a database.

[1027] Step 3:

[1028] When the match starts, the server notifies the AI ​​models of the set tasks. The input is detailed task information (such as a prompt or the URL of the dataset), and the output is a task notification to each participating AI model. Specifically, the server sends a notification containing task details to each AI model.

[1029] Step 4:

[1030] Each AI model generates an answer to the task and sends it to the server. The input is detailed information about the task, and the output is the generated answer. The AI ​​model analyzes the received task prompt and processes it to generate an answer.

[1031] Step 5:

[1032] The server scores the received answers based on the evaluation criteria. The input is the answer from each AI model and the pre-defined evaluation criteria, and the output is the score of the answer. The server analyzes the answers according to the evaluation criteria and assigns a score to each answer.

[1033] Step 6:

[1034] The server records the match results in a database and publishes them on a web interface. The input is the match score and answer data, and the output is the published match results. The server stores the match results, scores of each AI model, generated answers, evaluation comments, etc. in a database and makes them accessible to users and developers.

[1035] Step 7:

[1036] The server uses an emotion engine to recognize the user's emotions. The input is the user's facial expression, tone of voice, and input data, and the output is the user's emotional state. Specifically, the server analyzes data from the camera and microphone using the emotion engine to identify the user's emotions.

[1037] Step 8:

[1038] The server adjusts the interface based on the user's emotional state recognized by the emotion engine. The input is the recognized emotional state, and the output is an adjusted interface. For example, if the user is happy, the color tone of the interface or the timing of information presentation can be changed.

[1039] Step 9:

[1040] The terminal (for developers) submits its own AI model to the server. The input is the AI ​​model file created by the developer, and the output is the AI ​​model registered on the server. The terminal uploads the AI ​​model and enters related information through a web interface.

[1041] Step 10:

[1042] Users can check the match results and perform detailed analysis. The input is the match result data published on the server, and the output is the information analyzed by the user. Specifically, users can view the match results and evaluation comments using a web interface.

[1043] Step 11:

[1044] The server selects the optimal answer by having the generative AI models compete with each other. The input is the answers from multiple generative AI models, and the output is the selected optimal answer. The server compares the answers from multiple generative AI models and presents the answer with the highest evaluation to the user.

[1045] This trend will enable the competitive environment between generative AI models and interface adjustments based on user emotions.

[1046] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1047] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1048] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[1049] [Third embodiment]

[1050] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1051] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[1052] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1053] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[1054] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1055] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1056] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1057] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1058] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1059] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1060] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1061] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[1062] This system provides an environment in which generative AIs tackle problems and compete with each other to find the best solutions. The system mainly consists of a server, a terminal (for developers), and a user. The specific functions and operations of each component are explained below.

[1063] server

[1064] 1. Registering an AI model

[1065] The server receives the AI ​​model uploaded from the developer's device. After receiving it, it assigns a unique ID to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. It also performs a security check to ensure that the model does not contain any malicious programs.

[1066] 2. Match Scheduling

[1067] The server schedules the next match based on a pre-defined schedule, selects which AI models will participate, and records match details (e.g., task content, evaluation criteria) in a database.

[1068] 3. Match execution

[1069] When the match starts, the server notifies the AI ​​models of the challenge. The challenge notification includes details such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on the evaluation criteria.

[1070] 4. Recording and publishing match results

[1071] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[1072] Terminal (for developers)

[1073] 1. Submitting an AI model

[1074] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[1075] 2. Check the match results

[1076] Developers can view their model's scores and detailed feedback through a match results interface, which allows them to improve their AI model for the next match.

[1077] User

[1078] 1. Watching a game

[1079] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[1080] 2. Analysis of results

[1081] After the match, users can analyze the results in detail, understanding how each AI model solved the task and how the score based on the evaluation criteria was determined.

[1082] Specific examples

[1083] Example 1: Text generation task

[1084] The server sets the task of "generating a poem." Developer A submits AI model A, and developer B submits AI model B. When the match begins, the server displays a prompt to "generate a poem with an autumn landscape theme." Each AI model generates a poem and submits it to the server. The server evaluates the generated poems and assigns them scores based on criteria such as literary merit and grammatical accuracy.

[1085] Example 2: Image classification task

[1086] The server sets the task of "classifying cat images." Developer C submits AI model C, and developer D submits AI model D. When the match begins, the server notifies the AI ​​model of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server scores the AI ​​models based on accuracy rate and classification speed.

[1087] In this way, the AI ​​Battlefield system provides an environment in which developers can compete against other AI models and a means for users to observe the progress of AI technology in real time.

[1088] The processing flow will be explained below.

[1089] Step 1:

[1090] Uploading an AI model

[1091] Device: Developers upload the AI ​​model from their own device via the web interface, select the model from the file selection screen, and press the send button.

[1092] Step 2:

[1093] Security checks for AI models

[1094] Server: Receives the uploaded AI model and performs security checks to ensure it does not contain malicious code or viruses.

[1095] Step 3:

[1096] Registering an AI model

[1097] Server: After security checks are complete, it assigns a unique ID to the AI ​​model and stores the model and its metadata (developer name, model name, version, etc.) in a database.

[1098] Step 4:

[1099] Match Scheduling

[1100] Server: Schedules the next match based on a pre-set schedule. At this time, it selects the participating AI models and records the match details (task content, evaluation criteria) in a database.

[1101] Step 5:

[1102] Assignment notifications

[1103] Server: When the match start time arrives, the server notifies the participating AI models of the challenge, including details such as the prompt and the dataset URL.

[1104] Step 6:

[1105] Generate answers

[1106] Terminal: Each participating AI model generates an answer to the task and sends the result to the server. The answer generation must be completed within a specified time.

[1107] Step 7:

[1108] Receiving answers

[1109] Server: Receives the answers sent from each AI model and stores them in a database. Monitors the reception status and confirms that all answers have been received.

[1110] Step 8:

[1111] Scoring answers

[1112] Server: Scores the received answers based on evaluation criteria, such as originality, accuracy, and efficiency. Stores the scoring results in a database.

[1113] Step 9:

[1114] Recording match results

[1115] Server: Generates match results based on the scoring results and records them in a database, including the scores of each AI model, generated answers, and evaluation comments.

[1116] Step 10:

[1117] Publication of match results

[1118] Server: The server publishes the match results on a web interface accessible to developers and users. The results page displays the scores and evaluation comments for each AI model.

[1119] Step 11:

[1120] Check the match results

[1121] Device: Developers can view match results, scores, and detailed feedback from their own devices through a web interface, which can be used to improve the game for the next time.

[1122] Step 12:

[1123] Watching a game

[1124] Users: Users can watch the matches in real time and observe the performance of each AI model, and after the match, they can also perform detailed analysis of the results.

[1125] Example 1

[1126] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1127] Conventional competition systems using AI models lacked an efficient workflow, from model registration to running matches, publishing results, and analyzing them. Furthermore, they often lacked important features such as security checks and detailed feedback, resulting in issues of low usability and reliability for AI developers. There was a need for an effective system that could solve these issues, promote competition between AI models, and present technological progress to users in real time.

[1128] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1129] In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for performing security checks, means for assigning unique identifiers to AI models, means for users to watch matches in real time through a web interface, means for analyzing match results in detail, means for providing detailed feedback on generated answers, and means for providing developers with information to improve their AI models. This realizes an environment in which a series of processes, from registering AI models to detailed analysis of match results, can be executed efficiently and reliably, and makes it possible to provide AI developers with useful information.

[1130] An "AI model" is software designed to perform a specific task using artificial intelligence techniques.

[1131] A "server" refers to a computer system that provides services to other devices over a network.

[1132] "Device" means an electronic device used by a user or developer to access the Service.

[1133] "User" means any person or organization that uses the system or service.

[1134] A "challenge" is a task or problem that an AI model is given to solve.

[1135] An "answer" is the result or response that an AI model generates based on a task.

[1136] "Evaluation criteria" refers to the criteria or scale used to score the generated answers.

[1137] "Scoring" refers to assigning points to generated answers based on evaluation criteria.

[1138] A "database" refers to a system for efficiently storing, managing, and retrieving information.

[1139] A "security check" is an inspection to ensure that a system or software does not contain malicious programs.

[1140] A "unique identifier" is a number or code assigned to an element to uniquely identify it from all other elements.

[1141] "Web interface" refers to a web page or application that can be accessed through a browser.

[1142] "Feedback" refers to evaluations and comments on the AI ​​model's answers, which can be used to improve the model.

[1143] "Developers" refers to the engineers and programmers who create AI models and submit them to the system.

[1144] "Real-time" means that an event is processed and displayed immediately, with almost no delay, after it occurs.

[1145] "Registration" refers to the act of adding new information or models to the system.

[1146] "Scheduling" refers to setting a timetable for the execution of a particular event or task.

[1147] This invention is a system that provides an environment in which generative AI models tackle problems and compete with each other to find the best solutions. This system is mainly composed of a server, a terminal (for developers), and a user.

[1148] server

[1149] The server plays a central role in managing and operating the entire system using multiple pieces of hardware and software. The main functions of the server are as follows:

[1150] Registering an AI model

[1151] The server receives the AI ​​model uploaded from the developer's device and assigns it a unique identifier. The received AI model and related information (developer name, model name, version, etc.) are stored in a database. A security check is performed to ensure that the model does not contain any malicious programs.

[1152] Match Scheduling

[1153] The server schedules the next match based on a pre-defined schedule, selects the participating AI models, and records match details (e.g., task content, evaluation criteria) in a database.

[1154] Match execution

[1155] When the match starts, the server notifies the AI ​​models of the challenge. This challenge notification includes detailed information such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on the evaluation criteria.

[1156] Recording and publishing match results

[1157] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[1158] As a concrete example, the server can set a challenge to "generate a poem." For example, developer A submits AI model A, and developer B submits AI model B. When the match starts, the server notifies the participants with a prompt to "generate a poem on the theme of autumn scenery." Each AI model generates a poem and submits it to the server. The server evaluates the generated poems and assigns them scores based on criteria such as literary merit and grammatical accuracy.

[1159] Terminal (for developers)

[1160] The terminals are used by developers to submit AI models and check match results. Developers can use their own terminals to submit AI models to the server through a web interface and check match results.

[1161] Submitting an AI model

[1162] Developers can submit AI models to the server via a web interface using their own devices, and can enter model summary and version information when submitting.

[1163] Check the match results

[1164] Developers can view match results through a web interface, view their model's scores and detailed feedback, and improve their AI models for the next match.

[1165] User

[1166] Users use the system primarily to watch games and analyze results.

[1167] Watching a game

[1168] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[1169] Analyzing the results

[1170] After the match, users can analyze the results in detail, understanding how each AI model solved the task and how the score based on the evaluation criteria was determined.

[1171] As a concrete example, if the server sets the task of "classifying images of cats," developer C submits AI model C, and developer D submits AI model D. When the time for the match to start arrives, the server notifies it of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server then scores them based on accuracy rate and classification speed.

[1172] Examples of prompts include "Generate a poem on the theme of autumn scenery" and "Determine whether the image below is a cat."

[1173] The system promotes competition between AI models and provides an effective environment where developers and users can obtain information tailored to their respective goals.

[1174] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1175] Step 1: Registering an AI model

[1176] The server receives the AI ​​model from the developer's device. Specifically, the developer uploads the AI ​​model file through a web interface, and the server receives it. The input data is the AI ​​model file, and the output is the model's unique identifier and registration results. After receiving it, the server assigns a unique identifier to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. In addition, it performs security checks to ensure the AI ​​model does not contain malicious programs. Security checks include virus scans and static code analysis.

[1177] Step 2: Scheduling matches

[1178] The server schedules the next match based on a pre-set schedule. The input data is a list of AI models in the database and schedule information, and the output is detailed match information. Specifically, the server selects models to participate in the match from the list of AI models stored in the database. Past performance and random selection are used as criteria. Match details include task content and evaluation criteria, and this information is recorded in the database.

[1179] Step 3: Assignment notification

[1180] When the match starts, the server notifies the AI ​​models of the task. The input data is the task content and a list of participating AI models, and the output is the notification result. The server notifies each AI model of detailed information such as the prompt text and the URL of the dataset. Specifically, the server communicates the task to each model using an API call.

[1181] Step 4: Generate and submit answers

[1182] Each AI model generates an answer to the task and sends it to the server. The input data is the task notification content, and the output is the generated answer. Specifically, the AI ​​model generates an answer based on the received prompt. For example, it generates a poem in response to the prompt "Generate a poem on the theme of autumn scenery." The generated answer is then sent to the server.

[1183] Step 5: Receiving and evaluating answers

[1184] The server receives the answers from the AI ​​model and scores them based on evaluation criteria. The input data is the received answers, and the output is the evaluation results and score. The server scores the received answers based on evaluation criteria such as literary value and grammatical accuracy.

[1185] Step 6: Record and publish match results

[1186] The server records the match results in a database and makes them publicly available through a web interface. The input data are the evaluation results and scores, and the output is the publicly available information. The server stores the scores, generated answers, and evaluation comments of each AI model in a database and makes them publicly available through a web interface accessible to developers and users.

[1187] Step 7: Submit your AI model

[1188] Developers submit AI models through a web interface using their own devices. The input data is the AI ​​model file, its summary, and version information, and the output is the submission results. Developers select and upload the model file, enter the model summary and version information, and submit.

[1189] Step 8: Check the match results

[1190] Developers check the match results through a web interface. The input data is the match result information, and the output is the displayed score and feedback. Developers can check the scores and detailed feedback of their AI models and use them to improve their AI models for the next match.

[1191] Step 9: Watch the game

[1192] Users use an interface that allows them to watch the game in real time. The input data is game information updated in real time, and the output is the viewing experience. Users can observe live how each AI model tackles the problem and what solutions it generates.

[1193] Step 10: Analyze the results

[1194] Users can analyze the match results in detail after the match is over. The input data is the published match result information, and the output is the analysis results. Users can analyze the answers generated by each AI model and their scores based on the evaluation criteria to understand how they solved the problem. By referring to the evaluation comments, users can grasp the strengths and areas for improvement of the AI ​​model.

[1195] (Application example 1)

[1196] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1197] Previous systems utilizing generative AI models simply evaluated and compared the performance of AI models, lacking the ability to incorporate user participation and evaluation in real time. Furthermore, there were insufficient means for developers to receive detailed feedback, making it time-consuming and labor-intensive to improve AI models. Furthermore, there was a lack of platforms where users could evaluate and enjoy generated content, limiting opportunities for the general public to understand and enjoy technological advances in generative AI.

[1198] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1199] In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for users to view and evaluate generated content in real time, and means for developers to receive detailed feedback on the models. This allows the technology of generative AI models to be utilized while incorporating users' real-time evaluations, allowing developers to quickly improve their models based on detailed feedback. Furthermore, the server functions as a platform where general users can enjoy content created by generative AI, thereby promoting widespread understanding of advances in generative AI technology.

[1200] An "AI model" is an algorithm and database that uses artificial intelligence technology to generate answers and predictions for specific problems.

[1201] A "problem" is a specific problem or task that an AI model must address, and the resulting solutions are evaluated against it.

[1202] "Answer" refers to the output generated by an AI model based on a task.

[1203] "Users" are participants who use the system to view and evaluate generated content in real time.

[1204] "Developer" refers to a person or organization that develops an AI model and registers it in the system to test its performance.

[1205] "Means for notifying tasks" refers to a function that allows the server to communicate the tasks that need to be solved to the AI ​​model.

[1206] "Means for receiving answers" refers to the function for receiving answers generated from the AI ​​model on the server side.

[1207] "Means for scoring" refers to a function that assigns points to received answers based on evaluation criteria.

[1208] "Means for recording and publishing match results" refers to the ability to store the evaluation results of scored answers and display them in a form accessible to users and developers.

[1209] The "means for viewing and rating in real time" is a function that allows users to view generated content in real time and rate it on the spot.

[1210] "Means for receiving feedback" is a function that allows developers to receive detailed comments and suggestions for improvement regarding the AI ​​model's match results and user evaluations.

[1211] This invention is a system for pitting generative AI models against each other in specific challenges, allowing users to view and evaluate the generated content. Specifically, it consists of a server, a developer terminal, and a user interface.

[1212] server

[1213] 1. Registering the AI ​​model:

[1214] The server receives the AI ​​model provided by the developer, assigns a unique ID, and stores it in a database. It also performs security checks to ensure that no malicious programs are included. This process can be implemented using Python and Flask, and TensorFlow and SQLite are used to store the AI ​​model.

[1215] 2. Match Scheduling:

[1216] The server schedules matches, determines the next match time, and selects the AI ​​models that will participate. This information is recorded in a database and notifications are sent as needed. The scheduling logic is automatic, based on pre-defined times and conditions.

[1217] 3. Assignment Notification:

[1218] When the game starts, the AI ​​model is notified of the task. The task includes a prompt and the URL of the required dataset. For example, the prompt might be "Generate a poem based on an autumn landscape." This process is notified to the AI ​​model via an HTTP request.

[1219] 4. Receiving and Scoring Answers:

[1220] The answers generated by each AI model are sent to the server, where they are then scored based on evaluation criteria. Scoring evaluates the quality of the generated content and its suitability for the task. This process uses natural language processing (NLP) technology and evaluation algorithms.

[1221] 5. Recording and Publication of Match Results:

[1222] Match results are recorded in a database and made publicly accessible to users and developers, allowing users to rate the generated content and developers to receive feedback.

[1223] Developer Device

[1224] 1. Submitting an AI model:

[1225] Developers submit AI models to the server from their own devices, providing details such as the model name, version information, and developer details. Submissions are done through a web-based interface.

[1226] 2. Match result confirmation and feedback:

[1227] Developers have an interface that allows them to view match results and detailed evaluation feedback for each model, providing them with immediate information they need to improve their models.

[1228] User

[1229] 1. Real-time viewing and evaluation:

[1230] Users can watch the progress of the match in real time. The generated content is displayed in real time, and users can rate and comment on it on the spot. This function enables user-participation in evaluation.

[1231] 2. Analysis of match results:

[1232] After the match, users can analyze the evaluation results and scoring details of the generated content, helping them see how their own evaluations are reflected and understand the technological progress of the generating AI.

[1233] Specific examples

[1234] One weekend, the task is to "generate a poem with the theme of autumn scenery." Developer A's model generates a poem that "speaks of autumn, with red and yellow leaves fluttering in the air, and sunsets that color the sky. A poem that resonates with the heart, like a whisper of the wind." Meanwhile, Developer B's model generates a poem that "speaks of autumn scenery that evokes warm memories in the heart, with cold winds blowing." Each poem is evaluated by users and assigned a score.

[1235] Prompt Sentence Examples

[1236] "Generate a poem with the theme of autumn scenery. We're looking for poems that incorporate the beautiful colors, the sound of the wind, the glow of the sunset, etc."

[1237] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1238] Step 1: Registering an AI model

[1239] Input: Developers upload the AI ​​model file from their own device to the server.

[1240] Processing: The server receives the uploaded AI model, assigns a unique ID, and records related information such as the model name, developer name, and version in a database. It also performs security checks to ensure there are no malicious programs.

[1241] Output: The AI ​​model is saved in a database, along with the model's unique ID and related information.

[1242] Step 2: Scheduling the match

[1243] Input: The server plans the next match based on a predefined match schedule.

[1244] Processing: The server selects participating AI models according to the schedule and records the match date, time, task content, and evaluation criteria in a database.

[1245] Output: The next match information is saved in the database and ready to be notified.

[1246] Step 3: Issue notification

[1247] Input: As the start time of the match approaches, the server notifies the AI ​​model of the task.

[1248] Processing: The server sends each AI model a prompt and the URL of the dataset via an HTTP request. For example, the prompt might be "Generate a poem based on an autumn landscape."

[1249] Output: Each AI model receives the task information.

[1250] Step 4: Generate and submit your answers

[1251] Input: The AI ​​model generates an answer based on the received challenge.

[1252] Processing: The AI ​​model follows the prompt, uses natural language processing techniques to generate poetry or other content, and sends the answer to the server.

[1253] Output: The generated answer is sent to the server.

[1254] Step 5: Receiving and scoring answers

[1255] Input: The server receives the answers sent by each AI model.

[1256] Processing: The server collects the answers and scores them based on a set of criteria, including literary merit and grammatical accuracy.

[1257] Output: The answer and score of each AI model are recorded in a database.

[1258] Step 6: Record and publish match results

[1259] Input: After scoring is complete, the server compiles the match results.

[1260] Processing: The server records the match results in a database and exposes them through an interface, making this information accessible to users and developers.

[1261] Output: The match results are published and available for users and developers to view.

[1262] Step 7: Real-time viewing and evaluation by users

[1263] Input: Users access the generated content to watch it in real time.

[1264] Processing: The server displays the generated content to the user in real time and provides a mechanism for the user to immediately submit ratings and comments.

[1265] Output: User ratings and comments are recorded in the system.

[1266] Step 8: Detailed analysis of match results

[1267] Input: After the match is over, users and developers have access to analyze the match results.

[1268] Processing: The server displays detailed match results and feedback, and provides tools that allow users to analyze the performance of each AI model.

[1269] Output: Detailed match results and analytical information are displayed to users and developers.

[1270] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1271] This system provides an environment in which generative AIs can tackle problems and compete with each other to find the best answers, and also combines an emotion engine that recognizes the user's emotions. The system mainly consists of a server, a terminal (for developers), and a user, and each component works in cooperation. The specific functions and operations of each component are explained below.

[1272] server

[1273] 1. Registering an AI model

[1274] The server receives the AI ​​model uploaded from the developer's device. After receiving it, the server assigns a unique ID to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. It also performs security checks to ensure that the model does not contain any malicious programs.

[1275] 2. Match Scheduling

[1276] The server schedules the next match based on a pre-defined schedule, selects which AI models will participate, and records match details (task content, evaluation criteria) in a database.

[1277] 3. Match execution

[1278] When the match starts, the server notifies the AI ​​models of the challenge. This notification includes details such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on the evaluation criteria.

[1279] 4. Recording and publishing match results

[1280] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[1281] 5. Emotion Engine

[1282] The server is equipped with an emotion engine that recognizes the user's emotions. This emotion engine analyzes the user's facial expressions, tone of voice, input data, etc. while watching a game or viewing game results, and recognizes the user's emotional state.

[1283] 6. Emotion-Based Interface Adjustment

[1284] The server dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine, for example, adjusting the interface's color scheme and the timing of information presentation if the user is excited.

[1285] Terminal (for developers)

[1286] 1. Submitting an AI model

[1287] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[1288] 2. Check the match results

[1289] Developers can view their model's scores and detailed feedback through a match results interface, which allows them to improve their AI model for the next match.

[1290] User

[1291] 1. Watching a game

[1292] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[1293] 2. Analysis of results

[1294] After a match, users can analyze the results in detail, understanding how each AI model solved the problem and how the score was determined based on the evaluation criteria. Furthermore, based on the emotional data recognized by the emotion engine, analysis and feedback are provided according to the user's emotional state.

[1295] Specific examples

[1296] Example 1: Text generation task

[1297] The server sets the task of "generating a poem." Developer A submits AI model A, and developer B submits AI model B. When the match begins, the server displays a prompt to "generate a poem with an autumn landscape theme." Each AI model generates a poem and sends it to the server. The server evaluates the generated poems and assigns them scores based on criteria such as literary value and grammatical accuracy. While watching the match, if the user's emotions are high, the interface changes to a brighter color tone.

[1298] Example 2: Image classification task

[1299] The server sets the task of "classifying cat images." Developer C submits AI model C, and developer D submits AI model D. When the match begins, the server notifies the user of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server scores the users based on accuracy rate and classification speed. When checking the match results, if the user is detected as confused, the interface displays additional support information appropriate to the situation.

[1300] In this way, the AI ​​Battlefield system not only provides a competitive environment between generative AI models, but also adjusts the interface to take user emotions into account, providing a more fulfilling experience.

[1301] The processing flow will be explained below.

[1302] MODE FOR CARRYING OUT THE INVENTION

[1303] Server Operation

[1304] Step 1:

[1305] Receiving AI model uploads

[1306] Server: Receives the AI ​​model uploaded from the developer's device. After receiving it, it starts a security check.

[1307] Step 2:

[1308] Security Check

[1309] Server: Performs security checks on the received AI model to check for malicious code and viruses.

[1310] Step 3:

[1311] Registering an AI model

[1312] Server: Once the security check is complete, it assigns a unique ID to the AI ​​model and stores the model and its metadata (developer name, model name, version) in a database.

[1313] Step 4:

[1314] Match Scheduling

[1315] Server: Schedules the next match based on a pre-set schedule, selects participating AI models, and records detailed match information (task content, evaluation criteria) in a database.

[1316] Step 5:

[1317] Assignment notifications

[1318] Server: When the match start time arrives, the server notifies the participating AI models of the challenge, including details such as the prompt and the dataset URL.

[1319] Step 6:

[1320] Receiving answers

[1321] Server: Receives answers from each participating AI model and stores them in a database. Monitors the reception status and confirms that all answers have been received.

[1322] Step 7:

[1323] Scoring

[1324] Server: Scores the received answers based on criteria such as originality, accuracy, and efficiency.

[1325] Step 8:

[1326] Recording the results

[1327] Server: Generates match results based on the scoring results and records them in a database, including the scores of each AI model, generated answers, and evaluation comments.

[1328] Step 9:

[1329] Publication of match results

[1330] Server: The server publishes the match results on a web interface accessible to developers and users. The results page displays the scores and evaluation comments for each AI model.

[1331] Emotion Engine Operation

[1332] Step 10:

[1333] User Emotion Recognition

[1334] Server: Uses an emotion engine to recognize emotions from the user's facial expressions, tone of voice, and input data.

[1335] Step 11:

[1336] Interface Adjustments

[1337] Server: Dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine. For example, if the user is excited, the color scheme of the interface or the timing of information presentation is adjusted.

[1338] User Actions

[1339] Step 12:

[1340] Watching a game

[1341] Users: Watch the match in real time and observe the performance of each AI model.

[1342] Step 13:

[1343] Analyzing the results

[1344] User: After the match, the results are analyzed in detail. Based on the emotional data recognized by the emotion engine, analysis and feedback are provided according to the user's emotional state.

[1345] Device (developer) operation

[1346] Step 14:

[1347] Submitting an AI model

[1348] Device: Developers submit AI models from their own devices through a web interface, entering model overview and version information.

[1349] Step 15:

[1350] Check the match results

[1351] On-device: Developers can view their model's scores and detailed feedback through an interface that allows them to check the match results and make adjustments to improve their model for next time.

[1352] Specific examples

[1353] Example 1: Text generation task

[1354] Step 1:

[1355] Assignment settings

[1356] Server: The server sets the task of "creating a poem."

[1357] Step 2:

[1358] Submitting an AI model

[1359] Terminal: Developer A submits AI model A, and developer B submits AI model B.

[1360] Step 3:

[1361] Assignment notifications

[1362] Server: When the match start time arrives, the server will announce the prompt to "Generate a poem with an autumn landscape theme."

[1363] Step 4:

[1364] Generating and receiving answers

[1365] Terminal: Each AI model generates a poem and sends it to the server, which receives the answer and evaluates it.

[1366] Step 5:

[1367] Scoring

[1368] Server: Evaluates the generated poems, scoring them on criteria such as literary merit and grammatical accuracy.

[1369] Step 6:

[1370] emotion recognition

[1371] Server: While watching the game, the emotion engine recognizes the user's emotions.

[1372] Step 7:

[1373] Interface Adjustments

[1374] Server: If the user is emotionally excited, change the interface color to a brighter tone.

[1375] Example 2: Image classification task

[1376] Step 1:

[1377] Assignment settings

[1378] Server: The server sets the task "classify images of cats."

[1379] Step 2:

[1380] Submitting an AI model

[1381] Terminal: Developer C submits AI model C, and developer D submits AI model D.

[1382] Step 3:

[1383] Assignment notifications

[1384] Server: When the match starts, a series of cat images will be displayed.

[1385] Step 4:

[1386] Generating and receiving answers

[1387] Terminal: Each AI model classifies an image and sends the results to the server, which receives and evaluates the answers.

[1388] Step 5:

[1389] Scoring

[1390] Server: Scoring is based on accuracy rate and classification speed.

[1391] Step 6:

[1392] emotion recognition

[1393] Server: When checking the match results, the emotion engine recognizes the user's emotions.

[1394] Step 7:

[1395] Interface Adjustments

[1396] Server: If the user is detected as confused, the interface displays additional support information.

[1397] In this way, the AI ​​Battlefield system provides a competitive environment for generative AI models and adjusts the interface to take user emotions into account, providing a more fulfilling experience.

[1398] Example 2

[1399] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1400] Conventional AI model competition systems simply record and publish the results of AI model competitions, without adjusting the interface to take into account the user's emotional state. This resulted in a lack of improvement in the user experience and reduced engagement with the system. Furthermore, security checks and scheduling of AI models were not performed efficiently, leaving the system lacking reliability and operational efficiency.

[1401] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for analyzing the user's emotional state, and means for dynamically adjusting the interface based on the user's emotional state. This makes it possible to provide the results of AI model matches in a real-time and reliable manner and to improve the user experience.

[1402] An "AI model" is an artificial intelligence software system that has been trained to perform a specific task.

[1403] A "match" is an event in which multiple AI models compete against each other based on a given task and their performance is evaluated.

[1404] A "problem" is a specific problem or requirement that the AI ​​model must solve, such as text generation or image classification.

[1405] "Notification" is an action in which the server communicates details of the problem to the AI ​​model.

[1406] An "answer" is the resulting data generated by an AI model based on a task.

[1407] "Evaluation criteria" are indicators and rules for evaluating answers generated by an AI model.

[1408] "Scoring" is the process of numerically evaluating an AI model's answer based on evaluation criteria.

[1409] "Match results" are data that include the evaluation results of each AI model's answers.

[1410] "Recording" is the action of saving the match results and related information in a database.

[1411] "Public" means providing match results for access by users and developers via a web interface or the like.

[1412] The "emotion engine" is a system that analyzes the user's facial expressions, tone of voice, input data, etc. to recognize their emotional state.

[1413] "Dynamic interface adjustment" refers to changing the appearance of a web interface and the timing of information presentation according to the user's emotional state.

[1414] This system provides an environment in which generative AI models can tackle problems and compete with each other to find the best solutions, and also combines an emotion engine that recognizes the user's emotions. The system mainly consists of a server, a terminal (for developers), and a user, and each component works in cooperation with each other.

[1415] server

[1416] Registering an AI model

[1417] The server receives the AI ​​model uploaded from the developer's device. After receiving it, the server assigns a unique ID to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. Databases used include MySQL and PostgreSQL. Security checks are also performed to ensure that the model does not contain malicious programs. The security checks use defined rules and a machine learning-based security engine.

[1418] Match Scheduling

[1419] The server schedules the next match using a Python script based on a pre-defined schedule. It selects which AI models will participate and records the match details (task content, evaluation criteria) in a database. The task content is selected from pre-defined prompts, dataset URLs, etc.

[1420] Match execution

[1421] When the match starts, the server uses an API to notify the AI ​​models of the challenge. This notification includes details such as the prompt and the dataset URL. Each AI model generates an answer to the challenge and sends it to the server. The server scores the received answers based on evaluation criteria. Scoring can be done using rule-based evaluation methods or machine learning models.

[1422] Recording and publishing match results

[1423] The server records the match results (scores, generated answers, and evaluation comments) in a database and publishes them on a web interface accessible to developers and users. This web interface is built using HTML / CSS and JavaScript.

[1424] Emotion Engine

[1425] The server is equipped with an emotion engine that recognizes the user's emotions. This emotion engine analyzes the user's facial expressions, tone of voice, input data, etc. while watching a game or viewing game results to recognize the user's emotional state. Examples of emotion engines used include Face++ and IBM Watson.

[1426] Emotion-based interface adjustment

[1427] The server dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine, for example, changing the interface's color scheme or the timing of information presentation if the user is excited.

[1428] Terminal (for developers)

[1429] Submitting an AI model

[1430] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[1431] Check the match results

[1432] Developers can view their model's scores and detailed feedback through a match results interface, which allows them to improve their AI models for the next match.

[1433] User

[1434] Watching a game

[1435] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and the solutions it generates.

[1436] Analyzing the results

[1437] After a match, users can analyze the results in detail, understanding how each AI model solved the problem and how the score was determined based on the evaluation criteria. Furthermore, based on the emotional data recognized by the emotion engine, analysis and feedback are provided according to the user's emotional state.

[1438] Specific examples

[1439] Example 1: Text generation task

[1440] The server sets the task of "generating a poem." For example, it sets a prompt such as "Generate a poem with an autumn landscape theme." Developer A submits AI model A, and developer B submits AI model B. When the match starts, the server notifies each AI model of the task. Each AI model generates a poem and sends it to the server. The server evaluates the generated poems and scores them based on criteria such as literary value and grammatical accuracy. While watching the match, if the user's emotions are high, the interface changes to a brighter color tone.

[1441] Example 2: Image classification task

[1442] The server sets the task of "classifying cat images." Developer C submits AI model C, and developer D submits AI model D. When the match begins, the server notifies each AI model of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server scores the models based on accuracy rate and classification speed. When checking the match results, if the interface recognizes that the user is confused, it displays support information appropriate to the situation.

[1443] In this way, the AI ​​model battle system not only provides a competitive environment between generated AI models, but also adjusts the interface to take the user's emotions into account, providing a more fulfilling experience.

[1444] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1445] Step 1:

[1446] The server receives the AI ​​model from the developer's device. During reception, communication encryption (e.g., HTTPS) is used to ensure data security. The input data is the AI ​​model file uploaded by the developer, and the output is a notification that reception is complete.

[1447] Step 2:

[1448] The server assigns a unique ID to the received AI model. This ID is used to identify the model within the system. The input data is an AI model file, and the output is an AI model with a unique ID.

[1449] Step 3:

[1450] The server stores the AI ​​model and related information (developer name, model name, version, etc.) in a database. The database used is MySQL, PostgreSQL, etc. The input data is the AI ​​model and its related information, and the output is the confirmation information stored in the database.

[1451] Step 4:

[1452] The server performs a security check on the received AI model to ensure it does not contain malicious code. The security engine performs the check using defined rules or machine learning-based methods. The input data is the AI ​​model file, and the output is the result of the security check.

[1453] Step 5:

[1454] The server schedules upcoming matches based on a pre-defined schedule using a Python script. The input data is the schedule information, and the output is the match schedule information.

[1455] Step 6:

[1456] The server retrieves a list of available AI models from the database and selects an AI model to participate in the next match randomly or according to pre-defined conditions. The input data is the stored AI model information, and the output is a list of selected AI models.

[1457] Step 7:

[1458] The server records the details of the match (task content, evaluation criteria, etc.) in a database. The input data is the setting information about the match, and the output is the detailed match information recorded in the database.

[1459] Step 8:

[1460] The server uses an API to notify each AI model of the challenge at the start of the match. This notification includes a prompt and a dataset URL. The input data are the prompt and dataset URL, and the output is confirmation that the notification was received.

[1461] Step 9:

[1462] Each AI model generates a solution to a problem and sends it to the server. The input data is the problem prompt or dataset, and the output is the generated answer.

[1463] Step 10:

[1464] The server scores the received answers based on the evaluation criteria. Scoring can be done using rule-based evaluation methods or machine learning models. The input data are the generated answers, and the output is the scored results.

[1465] Step 11:

[1466] The server records the match results (e.g., scores, generated answers, and evaluation comments) in a database. The input data are the scoring results, and the output is the match results stored in the database.

[1467] Step 12:

[1468] The server publishes the match results through a web interface built using HTML / CSS and JavaScript. The input data are the match results from the database, and the output is the published match results page.

[1469] Step 13:

[1470] The server collects the user's facial expressions, tone of voice, input data, etc. and temporarily stores them in storage. The input data is the emotional state data from the user, and the output is the data stored in storage.

[1471] Step 14:

[1472] The server uses an emotion engine (e.g., Face++ or IBM Watson) to analyze the collected user data and recognize the emotional state. The input data is the emotional state data, and the output is the recognized emotional state.

[1473] Step 15:

[1474] The server dynamically adjusts the web interface based on the recognized emotional state, for example, changing the interface's color scheme or the timing of information presentation if the user is excited. The input data is the recognized emotional state, and the output is the dynamically adjusted interface.

[1475] Through this processing step, the AI ​​model competition system provides a competitive environment for generated AI models and also adjusts the interface to take user emotions into account, providing a more fulfilling experience.

[1476] (Application example 2)

[1477] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1478] Conventional competition systems using generative AI models lack the ability to adjust the interface to take into account user emotions and real-time feedback, resulting in a limited user experience. Furthermore, there is a lack of a means to visualize the competition between generative AI models and effectively communicate the results to users. Specifically, there is a need for a system that can change the interface based on emotional states, recommend appropriate products, and provide these in an integrated manner.

[1479] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for recognizing user emotions, means for adjusting the interface based on the user's emotional state, and means for having generated AI models compete with each other to select the optimal answer. This makes it possible to adjust the interface according to the user's emotional state and recommend optimal products.

[1480] An "AI model" is a collection of artificial intelligence algorithms designed to solve a specific problem.

[1481] A "means for scheduling a match" is a method for determining the date and time of the next match and planning the match for the AI ​​model based on a pre-set schedule.

[1482] The "means of notifying the task" refers to the method of transmitting detailed information about the task to the AI ​​model.

[1483] The "means of receiving the answer" is the mechanism by which the AI ​​model receives the answer generated based on the task.

[1484] A "criteria-based scoring means" is a method of evaluating generated answers according to predetermined criteria and assigning points.

[1485] "Means for recording and publishing match results" means a method for storing match results and publishing them in a form accessible to interested parties.

[1486] "Means for recognizing user emotions" refers to a method for identifying the emotions a user is feeling by analyzing the user's facial expressions, tone of voice, input data, etc.

[1487] "Means for adjusting the interface" refers to a method for dynamically changing the color tone and display content of the screen based on the user's emotional state.

[1488] "Method of having generative AI models compete with each other to select the optimal answer" is a method in which multiple generative AI models tackle the same task and select the best answer from among them.

[1489] This invention is a system that provides an environment in which generative AI models tackle problems and compete with each other for their solutions, and also combines it with an emotion engine that recognizes the user's emotions. This system is mainly composed of a server, a terminal (for developers), and a user, and each component works in cooperation with each other.

[1490] server

[1491] 1. Registering an AI model

[1492] The server receives the AI ​​model uploaded by the developer from the device. After receiving it, it assigns a unique ID and stores the model and related information (developer name, model name, version, etc.) in a database. It also performs a security check to ensure that the model does not contain any malicious programs.

[1493] 2. Match Scheduling

[1494] The server schedules matches based on a pre-set schedule, selects which AI models will participate, and records the tasks and evaluation criteria in a database.

[1495] 3. Match execution

[1496] When the match starts, the server notifies the AI ​​models of the challenge. This notification includes details such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on a set of criteria.

[1497] 4. Recording and publishing match results

[1498] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[1499] 5. Emotion Engine

[1500] The server is equipped with an emotion engine that recognizes the user's emotions. This emotion engine analyzes the user's facial expressions, tone of voice, and input data while watching a game or viewing game results, and recognizes the user's emotional state.

[1501] 6. Emotion-Based Interface Adjustment

[1502] The server dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine, for example, adjusting the interface's color scheme or the timing of information presentation if the user is excited.

[1503] Terminal (for developers)

[1504] 1. Submitting an AI model

[1505] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[1506] 2. Check the match results

[1507] Developers can view their model's scores and detailed feedback through a match result interface, which allows them to improve their AI model for the next match.

[1508] User

[1509] 1. Watching a game

[1510] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[1511] 2. Analysis of results

[1512] After a match, users can analyze the results in detail, understanding how each AI model solved the problem and how the score was determined based on the evaluation criteria. Furthermore, analysis and feedback based on the user's emotional state are provided based on emotional data recognized by the emotion engine.

[1513] Specific examples

[1514] Example 1: Emotion-based product recommendations

[1515] When a user is using the virtual store, their facial expressions are detected in real time via a camera. If the user is determined to be happy, the interface is adjusted to a brighter color tone and products with many positive reviews are recommended. At the same time, Generative AI Model A (specializing in fashion) and Generative AI Model B (specializing in gadgets) compete to recommend the best products for the user.

[1516] Example

[1517] User information: User's age, gender, purchase history

[1518] Camera frame: User's face image

[1519] Prompt statement:

[1520] User Information

[1521] user_profile = {

[1522] 'age': 30,

[1523] 'gender': 'male',

[1524] 'purchase_history': ['laptop', 'smartphone']

[1525] }

[1526] Frames from the camera

[1527] video_frame = cv2.imread('user_face.jpg')

[1528] recommendations, theme = main(user_profile, video_frame)

[1529] print(f'Recommendations: {recommendations}')

[1530] print(f'Interface Theme: {theme}')

[1531] By utilizing this invention, it becomes possible to flexibly adjust the interface according to the user's emotional state and to achieve highly accurate product recommendations through competition between generative AI models.

[1532] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1533] Step 1:

[1534] The server receives AI models uploaded from the device. The input is the AI ​​model file, and the output is a database that stores the model and related information (developer name, model name, version, etc.). Specifically, the server assigns a unique ID to the received AI model and performs security checks to ensure that it does not contain any malicious code.

[1535] Step 2:

[1536] The server schedules matches based on a pre-set schedule. The input is schedule information and a list of registered AI models, and the output is the date and time of the next match, the participating AI models, and a record of the task content. The server selects which AI models will participate and records the task content and evaluation criteria in a database.

[1537] Step 3:

[1538] When the match starts, the server notifies the AI ​​models of the set tasks. The input is detailed task information (such as a prompt or the URL of the dataset), and the output is a task notification to each participating AI model. Specifically, the server sends a notification containing task details to each AI model.

[1539] Step 4:

[1540] Each AI model generates an answer to the task and sends it to the server. The input is detailed information about the task, and the output is the generated answer. The AI ​​model analyzes the received task prompt and processes it to generate an answer.

[1541] Step 5:

[1542] The server scores the received answers based on the evaluation criteria. The input is the answer from each AI model and the pre-defined evaluation criteria, and the output is the score of the answer. The server analyzes the answers according to the evaluation criteria and assigns a score to each answer.

[1543] Step 6:

[1544] The server records the match results in a database and publishes them on a web interface. The input is the match score and answer data, and the output is the published match results. The server stores the match results, scores of each AI model, generated answers, evaluation comments, etc. in a database and makes them accessible to users and developers.

[1545] Step 7:

[1546] The server uses an emotion engine to recognize the user's emotions. The input is the user's facial expression, tone of voice, and input data, and the output is the user's emotional state. Specifically, the server analyzes data from the camera and microphone using the emotion engine to identify the user's emotions.

[1547] Step 8:

[1548] The server adjusts the interface based on the user's emotional state recognized by the emotion engine. The input is the recognized emotional state, and the output is an adjusted interface. For example, if the user is happy, the color tone of the interface or the timing of information presentation can be changed.

[1549] Step 9:

[1550] The terminal (for developers) submits its own AI model to the server. The input is the AI ​​model file created by the developer, and the output is the AI ​​model registered on the server. The terminal uploads the AI ​​model and enters related information through a web interface.

[1551] Step 10:

[1552] Users can check the match results and perform detailed analysis. The input is the match result data published on the server, and the output is the information analyzed by the user. Specifically, users can view the match results and evaluation comments using a web interface.

[1553] Step 11:

[1554] The server selects the optimal answer by having the generative AI models compete with each other. The input is the answers from multiple generative AI models, and the output is the selected optimal answer. The server compares the answers from multiple generative AI models and presents the answer with the highest evaluation to the user.

[1555] This trend will enable the competitive environment between generative AI models and interface adjustments based on user emotions.

[1556] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1557] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1558] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1559] [Fourth embodiment]

[1560] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1561] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1562] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1563] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1564] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1565] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1566] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1567] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1568] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1569] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1570] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1571] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1572] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1573] This system provides an environment in which generative AIs tackle problems and compete with each other to find the best solutions. The system mainly consists of a server, a terminal (for developers), and a user. The specific functions and operations of each component are explained below.

[1574] server

[1575] 1. Registering an AI model

[1576] The server receives the AI ​​model uploaded from the developer's device. After receiving it, it assigns a unique ID to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. It also performs a security check to ensure that the model does not contain any malicious programs.

[1577] 2. Match Scheduling

[1578] The server schedules the next match based on a pre-defined schedule, selects which AI models will participate, and records match details (e.g., task content, evaluation criteria) in a database.

[1579] 3. Match execution

[1580] When the match starts, the server notifies the AI ​​models of the challenge. The challenge notification includes details such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on the evaluation criteria.

[1581] 4. Recording and publishing match results

[1582] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[1583] Terminal (for developers)

[1584] 1. Submitting an AI model

[1585] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[1586] 2. Check the match results

[1587] Developers can view their model's scores and detailed feedback through a match results interface, which allows them to improve their AI model for the next match.

[1588] User

[1589] 1. Watching a game

[1590] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[1591] 2. Analysis of results

[1592] After the match, users can analyze the results in detail, understanding how each AI model solved the task and how the score based on the evaluation criteria was determined.

[1593] Specific examples

[1594] Example 1: Text generation task

[1595] The server sets the task of "generating a poem." Developer A submits AI model A, and developer B submits AI model B. When the match begins, the server displays a prompt to "generate a poem with an autumn landscape theme." Each AI model generates a poem and submits it to the server. The server evaluates the generated poems and assigns them scores based on criteria such as literary merit and grammatical accuracy.

[1596] Example 2: Image classification task

[1597] The server sets the task of "classifying cat images." Developer C submits AI model C, and developer D submits AI model D. When the match begins, the server notifies the AI ​​model of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server scores the AI ​​models based on accuracy rate and classification speed.

[1598] In this way, the AI ​​Battlefield system provides an environment in which developers can compete against other AI models and a means for users to observe the progress of AI technology in real time.

[1599] The processing flow will be explained below.

[1600] Step 1:

[1601] Uploading an AI model

[1602] Device: Developers upload the AI ​​model from their own device via the web interface, select the model from the file selection screen, and press the send button.

[1603] Step 2:

[1604] Security checks for AI models

[1605] Server: Receives the uploaded AI model and performs security checks to ensure it does not contain malicious code or viruses.

[1606] Step 3:

[1607] Registering an AI model

[1608] Server: After security checks are complete, it assigns a unique ID to the AI ​​model and stores the model and its metadata (developer name, model name, version, etc.) in a database.

[1609] Step 4:

[1610] Match Scheduling

[1611] Server: Schedules the next match based on a pre-set schedule. At this time, it selects the participating AI models and records the match details (task content, evaluation criteria) in a database.

[1612] Step 5:

[1613] Assignment notifications

[1614] Server: When the match start time arrives, the server notifies the participating AI models of the challenge, including details such as the prompt and the dataset URL.

[1615] Step 6:

[1616] Generate answers

[1617] Terminal: Each participating AI model generates an answer to the task and sends the result to the server. The answer generation must be completed within a specified time.

[1618] Step 7:

[1619] Receiving answers

[1620] Server: Receives the answers sent from each AI model and stores them in a database. Monitors the reception status and confirms that all answers have been received.

[1621] Step 8:

[1622] Scoring answers

[1623] Server: Scores the received answers based on evaluation criteria, such as originality, accuracy, and efficiency. Stores the scoring results in a database.

[1624] Step 9:

[1625] Recording match results

[1626] Server: Generates match results based on the scoring results and records them in a database, including the scores of each AI model, generated answers, and evaluation comments.

[1627] Step 10:

[1628] Publication of match results

[1629] Server: The server publishes the match results on a web interface accessible to developers and users. The results page displays the scores and evaluation comments for each AI model.

[1630] Step 11:

[1631] Check the match results

[1632] Device: Developers can view match results, scores, and detailed feedback from their own devices through a web interface, which can be used to improve the game for the next time.

[1633] Step 12:

[1634] Watching a game

[1635] Users: Users can watch the matches in real time and observe the performance of each AI model, and after the match, they can also perform detailed analysis of the results.

[1636] Example 1

[1637] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1638] Conventional competition systems using AI models lacked an efficient workflow, from model registration to running matches, publishing results, and analyzing them. Furthermore, they often lacked important features such as security checks and detailed feedback, resulting in issues of low usability and reliability for AI developers. There was a need for an effective system that could solve these issues, promote competition between AI models, and present technological progress to users in real time.

[1639] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1640] In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for performing security checks, means for assigning unique identifiers to AI models, means for users to watch matches in real time through a web interface, means for analyzing match results in detail, means for providing detailed feedback on generated answers, and means for providing developers with information to improve their AI models. This realizes an environment in which a series of processes, from registering AI models to detailed analysis of match results, can be executed efficiently and reliably, and makes it possible to provide AI developers with useful information.

[1641] An "AI model" is software designed to perform a specific task using artificial intelligence techniques.

[1642] A "server" refers to a computer system that provides services to other devices over a network.

[1643] "Device" means an electronic device used by a user or developer to access the Service.

[1644] "User" means any person or organization that uses the system or service.

[1645] A "challenge" is a task or problem that an AI model is given to solve.

[1646] An "answer" is the result or response that an AI model generates based on a task.

[1647] "Evaluation criteria" refers to the criteria or scale used to score the generated answers.

[1648] "Scoring" refers to assigning points to generated answers based on evaluation criteria.

[1649] A "database" refers to a system for efficiently storing, managing, and retrieving information.

[1650] A "security check" is an inspection to ensure that a system or software does not contain malicious programs.

[1651] A "unique identifier" is a number or code assigned to an element to uniquely identify it from all other elements.

[1652] "Web interface" refers to a web page or application that can be accessed through a browser.

[1653] "Feedback" refers to evaluations and comments on the AI ​​model's answers, which can be used to improve the model.

[1654] "Developers" refers to the engineers and programmers who create AI models and submit them to the system.

[1655] "Real-time" means that an event is processed and displayed immediately, with almost no delay, after it occurs.

[1656] "Registration" refers to the act of adding new information or models to the system.

[1657] "Scheduling" refers to setting a timetable for the execution of a particular event or task.

[1658] This invention is a system that provides an environment in which generative AI models tackle problems and compete with each other to find the best solutions. This system is mainly composed of a server, a terminal (for developers), and a user.

[1659] server

[1660] The server plays a central role in managing and operating the entire system using multiple pieces of hardware and software. The main functions of the server are as follows:

[1661] Registering an AI model

[1662] The server receives the AI ​​model uploaded from the developer's device and assigns it a unique identifier. The received AI model and related information (developer name, model name, version, etc.) are stored in a database. A security check is performed to ensure that the model does not contain any malicious programs.

[1663] Match Scheduling

[1664] The server schedules the next match based on a pre-defined schedule, selects the participating AI models, and records match details (e.g., task content, evaluation criteria) in a database.

[1665] Match execution

[1666] When the match starts, the server notifies the AI ​​models of the challenge. This challenge notification includes detailed information such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on the evaluation criteria.

[1667] Recording and publishing match results

[1668] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[1669] As a concrete example, the server can set a challenge to "generate a poem." For example, developer A submits AI model A, and developer B submits AI model B. When the match starts, the server notifies the participants with a prompt to "generate a poem on the theme of autumn scenery." Each AI model generates a poem and submits it to the server. The server evaluates the generated poems and assigns them scores based on criteria such as literary merit and grammatical accuracy.

[1670] Terminal (for developers)

[1671] The terminals are used by developers to submit AI models and check match results. Developers can use their own terminals to submit AI models to the server through a web interface and check match results.

[1672] Submitting an AI model

[1673] Developers can submit AI models to the server via a web interface using their own devices, and can enter model summary and version information when submitting.

[1674] Check the match results

[1675] Developers can view match results through a web interface, view their model's scores and detailed feedback, and improve their AI models for the next match.

[1676] User

[1677] Users use the system primarily to watch games and analyze results.

[1678] Watching a game

[1679] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[1680] Analyzing the results

[1681] After the match, users can analyze the results in detail, understanding how each AI model solved the task and how the score based on the evaluation criteria was determined.

[1682] As a concrete example, if the server sets the task of "classifying images of cats," developer C submits AI model C, and developer D submits AI model D. When the time for the match to start arrives, the server notifies it of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server then scores them based on accuracy rate and classification speed.

[1683] Examples of prompts include "Generate a poem on the theme of autumn scenery" and "Determine whether the image below is a cat."

[1684] The system promotes competition between AI models and provides an effective environment where developers and users can obtain information tailored to their respective goals.

[1685] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1686] Step 1: Registering an AI model

[1687] The server receives the AI ​​model from the developer's device. Specifically, the developer uploads the AI ​​model file through a web interface, and the server receives it. The input data is the AI ​​model file, and the output is the model's unique identifier and registration results. After receiving it, the server assigns a unique identifier to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. In addition, it performs security checks to ensure the AI ​​model does not contain malicious programs. Security checks include virus scans and static code analysis.

[1688] Step 2: Scheduling matches

[1689] The server schedules the next match based on a pre-set schedule. The input data is a list of AI models in the database and schedule information, and the output is detailed match information. Specifically, the server selects models to participate in the match from the list of AI models stored in the database. Past performance and random selection are used as criteria. Match details include task content and evaluation criteria, and this information is recorded in the database.

[1690] Step 3: Assignment notification

[1691] When the match starts, the server notifies the AI ​​models of the task. The input data is the task content and a list of participating AI models, and the output is the notification result. The server notifies each AI model of detailed information such as the prompt text and the URL of the dataset. Specifically, the server communicates the task to each model using an API call.

[1692] Step 4: Generate and submit answers

[1693] Each AI model generates an answer to the task and sends it to the server. The input data is the task notification content, and the output is the generated answer. Specifically, the AI ​​model generates an answer based on the received prompt. For example, it generates a poem in response to the prompt "Generate a poem on the theme of autumn scenery." The generated answer is then sent to the server.

[1694] Step 5: Receiving and evaluating answers

[1695] The server receives the answers from the AI ​​model and scores them based on evaluation criteria. The input data is the received answers, and the output is the evaluation results and score. The server scores the received answers based on evaluation criteria such as literary value and grammatical accuracy.

[1696] Step 6: Record and publish match results

[1697] The server records the match results in a database and makes them publicly available through a web interface. The input data are the evaluation results and scores, and the output is the publicly available information. The server stores the scores, generated answers, and evaluation comments of each AI model in a database and makes them publicly available through a web interface accessible to developers and users.

[1698] Step 7: Submit your AI model

[1699] Developers submit AI models through a web interface using their own devices. The input data is the AI ​​model file, its summary, and version information, and the output is the submission results. Developers select and upload the model file, enter the model summary and version information, and submit.

[1700] Step 8: Check the match results

[1701] Developers check the match results through a web interface. The input data is the match result information, and the output is the displayed score and feedback. Developers can check the scores and detailed feedback of their AI models and use them to improve their AI models for the next match.

[1702] Step 9: Watch the game

[1703] Users use an interface that allows them to watch the game in real time. The input data is game information updated in real time, and the output is the viewing experience. Users can observe live how each AI model tackles the problem and what solutions it generates.

[1704] Step 10: Analyze the results

[1705] Users can analyze the match results in detail after the match is over. The input data is the published match result information, and the output is the analysis results. Users can analyze the answers generated by each AI model and their scores based on the evaluation criteria to understand how they solved the problem. By referring to the evaluation comments, users can grasp the strengths and areas for improvement of the AI ​​model.

[1706] (Application example 1)

[1707] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1708] Previous systems utilizing generative AI models simply evaluated and compared the performance of AI models, lacking the ability to incorporate user participation and evaluation in real time. Furthermore, there were insufficient means for developers to receive detailed feedback, making it time-consuming and labor-intensive to improve AI models. Furthermore, there was a lack of platforms where users could evaluate and enjoy generated content, limiting opportunities for the general public to understand and enjoy technological advances in generative AI.

[1709] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1710] In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for users to view and evaluate generated content in real time, and means for developers to receive detailed feedback on the models. This allows the technology of generative AI models to be utilized while incorporating users' real-time evaluations, allowing developers to quickly improve their models based on detailed feedback. Furthermore, the server functions as a platform where general users can enjoy content created by generative AI, thereby promoting widespread understanding of advances in generative AI technology.

[1711] An "AI model" is an algorithm and database that uses artificial intelligence technology to generate answers and predictions for specific problems.

[1712] A "problem" is a specific problem or task that an AI model must address, and the resulting solutions are evaluated against it.

[1713] "Answer" refers to the output generated by an AI model based on a task.

[1714] "Users" are participants who use the system to view and evaluate generated content in real time.

[1715] "Developer" refers to a person or organization that develops an AI model and registers it in the system to test its performance.

[1716] "Means for notifying tasks" refers to a function that allows the server to communicate the tasks that need to be solved to the AI ​​model.

[1717] "Means for receiving answers" refers to the function for receiving answers generated from the AI ​​model on the server side.

[1718] "Means for scoring" refers to a function that assigns points to received answers based on evaluation criteria.

[1719] "Means for recording and publishing match results" refers to the ability to store the evaluation results of scored answers and display them in a form accessible to users and developers.

[1720] The "means for viewing and rating in real time" is a function that allows users to view generated content in real time and rate it on the spot.

[1721] "Means for receiving feedback" is a function that allows developers to receive detailed comments and suggestions for improvement regarding the AI ​​model's match results and user evaluations.

[1722] This invention is a system for pitting generative AI models against each other in specific challenges, allowing users to view and evaluate the generated content. Specifically, it consists of a server, a developer terminal, and a user interface.

[1723] server

[1724] 1. Registering the AI ​​model:

[1725] The server receives the AI ​​model provided by the developer, assigns a unique ID, and stores it in a database. It also performs security checks to ensure that no malicious programs are included. This process can be implemented using Python and Flask, and TensorFlow and SQLite are used to store the AI ​​model.

[1726] 2. Match Scheduling:

[1727] The server schedules matches, determines the next match time, and selects the AI ​​models that will participate. This information is recorded in a database and notifications are sent as needed. The scheduling logic is automatic, based on pre-defined times and conditions.

[1728] 3. Assignment Notification:

[1729] When the game starts, the AI ​​model is notified of the task. The task includes a prompt and the URL of the required dataset. For example, the prompt might be "Generate a poem based on an autumn landscape." This process is notified to the AI ​​model via an HTTP request.

[1730] 4. Receiving and Scoring Answers:

[1731] The answers generated by each AI model are sent to the server, where they are then scored based on evaluation criteria. Scoring evaluates the quality of the generated content and its suitability for the task. This process uses natural language processing (NLP) technology and evaluation algorithms.

[1732] 5. Recording and Publication of Match Results:

[1733] Match results are recorded in a database and made publicly accessible to users and developers, allowing users to rate the generated content and developers to receive feedback.

[1734] Developer Device

[1735] 1. Submitting an AI model:

[1736] Developers submit AI models to the server from their own devices, providing details such as the model name, version information, and developer details. Submissions are done through a web-based interface.

[1737] 2. Match result confirmation and feedback:

[1738] Developers have an interface that allows them to view match results and detailed evaluation feedback for each model, providing them with immediate information they need to improve their models.

[1739] User

[1740] 1. Real-time viewing and evaluation:

[1741] Users can watch the progress of the match in real time. The generated content is displayed in real time, and users can rate and comment on it on the spot. This function enables user-participation in evaluation.

[1742] 2. Analysis of match results:

[1743] After the match, users can analyze the evaluation results and scoring details of the generated content, helping them see how their own evaluations are reflected and understand the technological progress of the generating AI.

[1744] Specific examples

[1745] One weekend, the task is to "generate a poem with the theme of autumn scenery." Developer A's model generates a poem that "speaks of autumn, with red and yellow leaves fluttering in the air, and sunsets that color the sky. A poem that resonates with the heart, like a whisper of the wind." Meanwhile, Developer B's model generates a poem that "speaks of autumn scenery that evokes warm memories in the heart, with cold winds blowing." Each poem is evaluated by users and assigned a score.

[1746] Prompt Sentence Examples

[1747] "Generate a poem with the theme of autumn scenery. We're looking for poems that incorporate the beautiful colors, the sound of the wind, the glow of the sunset, etc."

[1748] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1749] Step 1: Registering an AI model

[1750] Input: Developers upload the AI ​​model file from their own device to the server.

[1751] Processing: The server receives the uploaded AI model, assigns a unique ID, and records related information such as the model name, developer name, and version in a database. It also performs security checks to ensure there are no malicious programs.

[1752] Output: The AI ​​model is saved in a database, along with the model's unique ID and related information.

[1753] Step 2: Scheduling the match

[1754] Input: The server plans the next match based on a predefined match schedule.

[1755] Processing: The server selects participating AI models according to the schedule and records the match date, time, task content, and evaluation criteria in a database.

[1756] Output: The next match information is saved in the database and ready to be notified.

[1757] Step 3: Issue notification

[1758] Input: As the start time of the match approaches, the server notifies the AI ​​model of the task.

[1759] Processing: The server sends each AI model a prompt and the URL of the dataset via an HTTP request. For example, the prompt might be "Generate a poem based on an autumn landscape."

[1760] Output: Each AI model receives the task information.

[1761] Step 4: Generate and submit your answers

[1762] Input: The AI ​​model generates an answer based on the received challenge.

[1763] Processing: The AI ​​model follows the prompt, uses natural language processing techniques to generate poetry or other content, and sends the answer to the server.

[1764] Output: The generated answer is sent to the server.

[1765] Step 5: Receiving and scoring answers

[1766] Input: The server receives the answers sent by each AI model.

[1767] Processing: The server collects the answers and scores them based on a set of criteria, including literary merit and grammatical accuracy.

[1768] Output: The answer and score of each AI model are recorded in a database.

[1769] Step 6: Record and publish match results

[1770] Input: After scoring is complete, the server compiles the match results.

[1771] Processing: The server records the match results in a database and exposes them through an interface, making this information accessible to users and developers.

[1772] Output: The match results are published and available for users and developers to view.

[1773] Step 7: Real-time viewing and evaluation by users

[1774] Input: Users access the generated content to watch it in real time.

[1775] Processing: The server displays the generated content to the user in real time and provides a mechanism for the user to immediately submit ratings and comments.

[1776] Output: User ratings and comments are recorded in the system.

[1777] Step 8: Detailed analysis of match results

[1778] Input: After the match is over, users and developers have access to analyze the match results.

[1779] Processing: The server displays detailed match results and feedback, and provides tools that allow users to analyze the performance of each AI model.

[1780] Output: Detailed match results and analytical information are displayed to users and developers.

[1781] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1782] This system provides an environment in which generative AIs can tackle problems and compete with each other to find the best answers, and also combines an emotion engine that recognizes the user's emotions. The system mainly consists of a server, a terminal (for developers), and a user, and each component works in cooperation. The specific functions and operations of each component are explained below.

[1783] server

[1784] 1. Registering an AI model

[1785] The server receives the AI ​​model uploaded from the developer's device. After receiving it, the server assigns a unique ID to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. It also performs security checks to ensure that the model does not contain any malicious programs.

[1786] 2. Match Scheduling

[1787] The server schedules the next match based on a pre-defined schedule, selects which AI models will participate, and records match details (task content, evaluation criteria) in a database.

[1788] 3. Match execution

[1789] When the match starts, the server notifies the AI ​​models of the challenge. This notification includes details such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on the evaluation criteria.

[1790] 4. Recording and publishing match results

[1791] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[1792] 5. Emotion Engine

[1793] The server is equipped with an emotion engine that recognizes the user's emotions. This emotion engine analyzes the user's facial expressions, tone of voice, input data, etc. while watching a game or viewing game results, and recognizes the user's emotional state.

[1794] 6. Emotion-Based Interface Adjustment

[1795] The server dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine, for example, adjusting the interface's color scheme and the timing of information presentation if the user is excited.

[1796] Terminal (for developers)

[1797] 1. Submitting an AI model

[1798] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[1799] 2. Check the match results

[1800] Developers can view their model's scores and detailed feedback through a match results interface, which allows them to improve their AI model for the next match.

[1801] User

[1802] 1. Watching a game

[1803] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[1804] 2. Analysis of results

[1805] After a match, users can analyze the results in detail, understanding how each AI model solved the problem and how the score was determined based on the evaluation criteria. Furthermore, based on the emotional data recognized by the emotion engine, analysis and feedback are provided according to the user's emotional state.

[1806] Specific examples

[1807] Example 1: Text generation task

[1808] The server sets the task of "generating a poem." Developer A submits AI model A, and developer B submits AI model B. When the match begins, the server displays a prompt to "generate a poem with an autumn landscape theme." Each AI model generates a poem and sends it to the server. The server evaluates the generated poems and assigns them scores based on criteria such as literary value and grammatical accuracy. While watching the match, if the user's emotions are high, the interface changes to a brighter color tone.

[1809] Example 2: Image classification task

[1810] The server sets the task of "classifying cat images." Developer C submits AI model C, and developer D submits AI model D. When the match begins, the server notifies the user of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server scores the users based on accuracy rate and classification speed. When checking the match results, if the user is detected as confused, the interface displays additional support information appropriate to the situation.

[1811] In this way, the AI ​​Battlefield system not only provides a competitive environment between generative AI models, but also adjusts the interface to take user emotions into account, providing a more fulfilling experience.

[1812] The processing flow will be explained below.

[1813] MODE FOR CARRYING OUT THE INVENTION

[1814] Server Operation

[1815] Step 1:

[1816] Receiving AI model uploads

[1817] Server: Receives the AI ​​model uploaded from the developer's device. After receiving it, it starts a security check.

[1818] Step 2:

[1819] Security Check

[1820] Server: Performs security checks on the received AI model to check for malicious code and viruses.

[1821] Step 3:

[1822] Registering an AI model

[1823] Server: Once the security check is complete, it assigns a unique ID to the AI ​​model and stores the model and its metadata (developer name, model name, version) in a database.

[1824] Step 4:

[1825] Match Scheduling

[1826] Server: Schedules the next match based on a pre-set schedule, selects participating AI models, and records detailed match information (task content, evaluation criteria) in a database.

[1827] Step 5:

[1828] Assignment notifications

[1829] Server: When the match start time arrives, the server notifies the participating AI models of the challenge, including details such as the prompt and the dataset URL.

[1830] Step 6:

[1831] Receiving answers

[1832] Server: Receives answers from each participating AI model and stores them in a database. Monitors the reception status and confirms that all answers have been received.

[1833] Step 7:

[1834] Scoring

[1835] Server: Scores the received answers based on criteria such as originality, accuracy, and efficiency.

[1836] Step 8:

[1837] Recording the results

[1838] Server: Generates match results based on the scoring results and records them in a database, including the scores of each AI model, generated answers, and evaluation comments.

[1839] Step 9:

[1840] Publication of match results

[1841] Server: The server publishes the match results on a web interface accessible to developers and users. The results page displays the scores and evaluation comments for each AI model.

[1842] Emotion Engine Operation

[1843] Step 10:

[1844] User Emotion Recognition

[1845] Server: Uses an emotion engine to recognize emotions from the user's facial expressions, tone of voice, and input data.

[1846] Step 11:

[1847] Interface Adjustments

[1848] Server: Dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine. For example, if the user is excited, the color scheme of the interface or the timing of information presentation is adjusted.

[1849] User Actions

[1850] Step 12:

[1851] Watching a game

[1852] Users: Watch the match in real time and observe the performance of each AI model.

[1853] Step 13:

[1854] Analyzing the results

[1855] User: After the match, the results are analyzed in detail. Based on the emotional data recognized by the emotion engine, analysis and feedback are provided according to the user's emotional state.

[1856] Device (developer) operation

[1857] Step 14:

[1858] Submitting an AI model

[1859] Device: Developers submit AI models from their own devices through a web interface, entering model overview and version information.

[1860] Step 15:

[1861] Check the match results

[1862] On-device: Developers can view their model's scores and detailed feedback through an interface that allows them to check the match results and make adjustments to improve their model for next time.

[1863] Specific examples

[1864] Example 1: Text generation task

[1865] Step 1:

[1866] Assignment settings

[1867] Server: The server sets the task of "creating a poem."

[1868] Step 2:

[1869] Submitting an AI model

[1870] Terminal: Developer A submits AI model A, and developer B submits AI model B.

[1871] Step 3:

[1872] Assignment notifications

[1873] Server: When the match start time arrives, the server will announce the prompt to "Generate a poem with an autumn landscape theme."

[1874] Step 4:

[1875] Generating and receiving answers

[1876] Terminal: Each AI model generates a poem and sends it to the server, which receives the answer and evaluates it.

[1877] Step 5:

[1878] Scoring

[1879] Server: Evaluates the generated poems, scoring them on criteria such as literary merit and grammatical accuracy.

[1880] Step 6:

[1881] emotion recognition

[1882] Server: While watching the game, the emotion engine recognizes the user's emotions.

[1883] Step 7:

[1884] Interface Adjustments

[1885] Server: If the user is emotionally excited, change the interface color to a brighter tone.

[1886] Example 2: Image classification task

[1887] Step 1:

[1888] Assignment settings

[1889] Server: The server sets the task "classify images of cats."

[1890] Step 2:

[1891] Submitting an AI model

[1892] Terminal: Developer C submits AI model C, and developer D submits AI model D.

[1893] Step 3:

[1894] Assignment notifications

[1895] Server: When the match starts, a series of cat images will be displayed.

[1896] Step 4:

[1897] Generating and receiving answers

[1898] Terminal: Each AI model classifies an image and sends the results to the server, which receives and evaluates the answers.

[1899] Step 5:

[1900] Scoring

[1901] Server: Scoring is based on accuracy rate and classification speed.

[1902] Step 6:

[1903] emotion recognition

[1904] Server: When checking the match results, the emotion engine recognizes the user's emotions.

[1905] Step 7:

[1906] Interface Adjustments

[1907] Server: If the user is detected as confused, the interface displays additional support information.

[1908] In this way, the AI ​​Battlefield system provides a competitive environment for generative AI models and adjusts the interface to take user emotions into account, providing a more fulfilling experience.

[1909] Example 2

[1910] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1911] Conventional AI model competition systems simply record and publish the results of AI model competitions, without adjusting the interface to take into account the user's emotional state. This resulted in a lack of improvement in the user experience and reduced engagement with the system. Furthermore, security checks and scheduling of AI models were not performed efficiently, leaving the system lacking reliability and operational efficiency.

[1912] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for analyzing the user's emotional state, and means for dynamically adjusting the interface based on the user's emotional state. This makes it possible to provide the results of AI model matches in a real-time and reliable manner and to improve the user experience.

[1913] An "AI model" is an artificial intelligence software system that has been trained to perform a specific task.

[1914] A "match" is an event in which multiple AI models compete against each other based on a given task and their performance is evaluated.

[1915] A "problem" is a specific problem or requirement that the AI ​​model must solve, such as text generation or image classification.

[1916] "Notification" is an action in which the server communicates details of the problem to the AI ​​model.

[1917] An "answer" is the resulting data generated by an AI model based on a task.

[1918] "Evaluation criteria" are indicators and rules for evaluating answers generated by an AI model.

[1919] "Scoring" is the process of numerically evaluating an AI model's answer based on evaluation criteria.

[1920] "Match results" are data that include the evaluation results of each AI model's answers.

[1921] "Recording" is the action of saving the match results and related information in a database.

[1922] "Public" means providing match results for access by users and developers via a web interface or the like.

[1923] The "emotion engine" is a system that analyzes the user's facial expressions, tone of voice, input data, etc. to recognize their emotional state.

[1924] "Dynamic interface adjustment" refers to changing the appearance of a web interface and the timing of information presentation according to the user's emotional state.

[1925] This system provides an environment in which generative AI models can tackle problems and compete with each other to find the best solutions, and also combines an emotion engine that recognizes the user's emotions. The system mainly consists of a server, a terminal (for developers), and a user, and each component works in cooperation with each other.

[1926] server

[1927] Registering an AI model

[1928] The server receives the AI ​​model uploaded from the developer's device. After receiving it, the server assigns a unique ID to the AI ​​model and stores the model and related information (developer name, model name, version, etc.) in a database. Databases used include MySQL and PostgreSQL. Security checks are also performed to ensure that the model does not contain malicious programs. The security checks use defined rules and a machine learning-based security engine.

[1929] Match Scheduling

[1930] The server schedules the next match using a Python script based on a pre-defined schedule. It selects which AI models will participate and records the match details (task content, evaluation criteria) in a database. The task content is selected from pre-defined prompts, dataset URLs, etc.

[1931] Match execution

[1932] When the match starts, the server uses an API to notify the AI ​​models of the challenge. This notification includes details such as the prompt and the dataset URL. Each AI model generates an answer to the challenge and sends it to the server. The server scores the received answers based on evaluation criteria. Scoring can be done using rule-based evaluation methods or machine learning models.

[1933] Recording and publishing match results

[1934] The server records the match results (scores, generated answers, and evaluation comments) in a database and publishes them on a web interface accessible to developers and users. This web interface is built using HTML / CSS and JavaScript.

[1935] Emotion Engine

[1936] The server is equipped with an emotion engine that recognizes the user's emotions. This emotion engine analyzes the user's facial expressions, tone of voice, input data, etc. while watching a game or viewing game results to recognize the user's emotional state. Examples of emotion engines used include Face++ and IBM Watson.

[1937] Emotion-based interface adjustment

[1938] The server dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine, for example, changing the interface's color scheme or the timing of information presentation if the user is excited.

[1939] Terminal (for developers)

[1940] Submitting an AI model

[1941] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[1942] Check the match results

[1943] Developers can view their model's scores and detailed feedback through a match results interface, which allows them to improve their AI models for the next match.

[1944] User

[1945] Watching a game

[1946] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and the solutions it generates.

[1947] Analyzing the results

[1948] After a match, users can analyze the results in detail, understanding how each AI model solved the problem and how the score was determined based on the evaluation criteria. Furthermore, based on the emotional data recognized by the emotion engine, analysis and feedback are provided according to the user's emotional state.

[1949] Specific examples

[1950] Example 1: Text generation task

[1951] The server sets the task of "generating a poem." For example, it sets a prompt such as "Generate a poem with an autumn landscape theme." Developer A submits AI model A, and developer B submits AI model B. When the match starts, the server notifies each AI model of the task. Each AI model generates a poem and sends it to the server. The server evaluates the generated poems and scores them based on criteria such as literary value and grammatical accuracy. While watching the match, if the user's emotions are high, the interface changes to a brighter color tone.

[1952] Example 2: Image classification task

[1953] The server sets the task of "classifying cat images." Developer C submits AI model C, and developer D submits AI model D. When the match begins, the server notifies each AI model of a series of cat images. Each AI model classifies the images as "cat" or "not cat" and sends the results to the server. The server scores the models based on accuracy rate and classification speed. When checking the match results, if the interface recognizes that the user is confused, it displays support information appropriate to the situation.

[1954] In this way, the AI ​​model battle system not only provides a competitive environment between generated AI models, but also adjusts the interface to take the user's emotions into account, providing a more fulfilling experience.

[1955] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1956] Step 1:

[1957] The server receives the AI ​​model from the developer's device. During reception, communication encryption (e.g., HTTPS) is used to ensure data security. The input data is the AI ​​model file uploaded by the developer, and the output is a notification that reception is complete.

[1958] Step 2:

[1959] The server assigns a unique ID to the received AI model. This ID is used to identify the model within the system. The input data is an AI model file, and the output is an AI model with a unique ID.

[1960] Step 3:

[1961] The server stores the AI ​​model and related information (developer name, model name, version, etc.) in a database. The database used is MySQL, PostgreSQL, etc. The input data is the AI ​​model and its related information, and the output is the confirmation information stored in the database.

[1962] Step 4:

[1963] The server performs a security check on the received AI model to ensure it does not contain malicious code. The security engine performs the check using defined rules or machine learning-based methods. The input data is the AI ​​model file, and the output is the result of the security check.

[1964] Step 5:

[1965] The server schedules upcoming matches based on a pre-defined schedule using a Python script. The input data is the schedule information, and the output is the match schedule information.

[1966] Step 6:

[1967] The server retrieves a list of available AI models from the database and selects an AI model to participate in the next match randomly or according to pre-defined conditions. The input data is the stored AI model information, and the output is a list of selected AI models.

[1968] Step 7:

[1969] The server records the details of the match (task content, evaluation criteria, etc.) in a database. The input data is the setting information about the match, and the output is the detailed match information recorded in the database.

[1970] Step 8:

[1971] The server uses an API to notify each AI model of the challenge at the start of the match. This notification includes a prompt and a dataset URL. The input data are the prompt and dataset URL, and the output is confirmation that the notification was received.

[1972] Step 9:

[1973] Each AI model generates a solution to a problem and sends it to the server. The input data is the problem prompt or dataset, and the output is the generated answer.

[1974] Step 10:

[1975] The server scores the received answers based on the evaluation criteria. Scoring can be done using rule-based evaluation methods or machine learning models. The input data are the generated answers, and the output is the scored results.

[1976] Step 11:

[1977] The server records the match results (e.g., scores, generated answers, and evaluation comments) in a database. The input data are the scoring results, and the output is the match results stored in the database.

[1978] Step 12:

[1979] The server publishes the match results through a web interface built using HTML / CSS and JavaScript. The input data are the match results from the database, and the output is the published match results page.

[1980] Step 13:

[1981] The server collects the user's facial expressions, tone of voice, input data, etc. and temporarily stores them in storage. The input data is the emotional state data from the user, and the output is the data stored in storage.

[1982] Step 14:

[1983] The server uses an emotion engine (e.g., Face++ or IBM Watson) to analyze the collected user data and recognize the emotional state. The input data is the emotional state data, and the output is the recognized emotional state.

[1984] Step 15:

[1985] The server dynamically adjusts the web interface based on the recognized emotional state, for example, changing the interface's color scheme or the timing of information presentation if the user is excited. The input data is the recognized emotional state, and the output is the dynamically adjusted interface.

[1986] Through this processing step, the AI ​​model competition system provides a competitive environment for generated AI models and also adjusts the interface to take user emotions into account, providing a more fulfilling experience.

[1987] (Application example 2)

[1988] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1989] Conventional competition systems using generative AI models lack the ability to adjust the interface to take into account user emotions and real-time feedback, resulting in a limited user experience. Furthermore, there is a lack of a means to visualize the competition between generative AI models and effectively communicate the results to users. Specifically, there is a need for a system that can change the interface based on emotional states, recommend appropriate products, and provide these in an integrated manner.

[1990] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for registering AI models, means for scheduling matches, means for notifying participating AI models of challenges, means for receiving answers generated by the AI ​​models based on the challenges, means for scoring the generated answers based on evaluation criteria, means for recording and publishing match results, means for recognizing user emotions, means for adjusting the interface based on the user's emotional state, and means for having generated AI models compete with each other to select the optimal answer. This makes it possible to adjust the interface according to the user's emotional state and recommend optimal products.

[1991] An "AI model" is a collection of artificial intelligence algorithms designed to solve a specific problem.

[1992] A "means for scheduling a match" is a method for determining the date and time of the next match and planning the match for the AI ​​model based on a pre-set schedule.

[1993] The "means of notifying the task" refers to the method of transmitting detailed information about the task to the AI ​​model.

[1994] The "means of receiving the answer" is the mechanism by which the AI ​​model receives the answer generated based on the task.

[1995] A "criteria-based scoring means" is a method of evaluating generated answers according to predetermined criteria and assigning points.

[1996] "Means for recording and publishing match results" means a method for storing match results and publishing them in a form accessible to interested parties.

[1997] "Means for recognizing user emotions" refers to a method for identifying the emotions a user is feeling by analyzing the user's facial expressions, tone of voice, input data, etc.

[1998] "Means for adjusting the interface" refers to a method for dynamically changing the color tone and display content of the screen based on the user's emotional state.

[1999] "Method of having generative AI models compete with each other to select the optimal answer" is a method in which multiple generative AI models tackle the same task and select the best answer from among them.

[2000] This invention is a system that provides an environment in which generative AI models tackle problems and compete with each other for their solutions, and also combines it with an emotion engine that recognizes the user's emotions. This system is mainly composed of a server, a terminal (for developers), and a user, and each component works in cooperation with each other.

[2001] server

[2002] 1. Registering an AI model

[2003] The server receives the AI ​​model uploaded by the developer from the device. After receiving it, it assigns a unique ID and stores the model and related information (developer name, model name, version, etc.) in a database. It also performs a security check to ensure that the model does not contain any malicious programs.

[2004] 2. Match Scheduling

[2005] The server schedules matches based on a pre-set schedule, selects which AI models will participate, and records the tasks and evaluation criteria in a database.

[2006] 3. Match execution

[2007] When the match starts, the server notifies the AI ​​models of the challenge. This notification includes details such as the prompt and the URL of the dataset. Each AI model generates an answer to the challenge and sends it to the server. The server receives the answers and scores each answer based on a set of criteria.

[2008] 4. Recording and publishing match results

[2009] The server records the match results in a database and publishes them to a web interface accessible to developers and users. The published information includes each AI model's score, generated answers, and evaluation comments.

[2010] 5. Emotion Engine

[2011] The server is equipped with an emotion engine that recognizes the user's emotions. This emotion engine analyzes the user's facial expressions, tone of voice, and input data while watching a game or viewing game results, and recognizes the user's emotional state.

[2012] 6. Emotion-Based Interface Adjustment

[2013] The server dynamically adjusts the web interface based on the user's emotional state as recognized by the emotion engine, for example, adjusting the interface's color scheme or the timing of information presentation if the user is excited.

[2014] Terminal (for developers)

[2015] 1. Submitting an AI model

[2016] Developers can submit AI models to the server via a web interface using their own devices, and can enter model overview and version information when submitting.

[2017] 2. Check the match results

[2018] Developers can view their model's scores and detailed feedback through a match result interface, which allows them to improve their AI model for the next match.

[2019] User

[2020] 1. Watching a game

[2021] Users can watch the game in real time through an interface, observing live how each AI model tackles the challenge and what solutions it generates.

[2022] 2. Analysis of results

[2023] After a match, users can analyze the results in detail, understanding how each AI model solved the problem and how the score was determined based on the evaluation criteria. Furthermore, analysis and feedback based on the user's emotional state are provided based on emotional data recognized by the emotion engine.

[2024] Specific examples

[2025] Example 1: Emotion-based product recommendations

[2026] When a user is using the virtual store, their facial expressions are detected in real time via a camera. If the user is determined to be happy, the interface is adjusted to a brighter color tone and products with many positive reviews are recommended. At the same time, Generative AI Model A (specializing in fashion) and Generative AI Model B (specializing in gadgets) compete to recommend the best products for the user.

[2027] Example

[2028] User information: User's age, gender, purchase history

[2029] Camera frame: User's face image

[2030] Prompt statement:

[2031] User Information

[2032] user_profile = {

[2033] 'age': 30,

[2034] 'gender': 'male',

[2035] 'purchase_history': ['laptop', 'smartphone']

[2036] }

[2037] Frames from the camera

[2038] video_frame = cv2.imread('user_face.jpg')

[2039] recommendations, theme = main(user_profile, video_frame)

[2040] print(f'Recommendations: {recommendations}')

[2041] print(f'Interface Theme: {theme}')

[2042] By utilizing this invention, it becomes possible to flexibly adjust the interface according to the user's emotional state and to achieve highly accurate product recommendations through competition between generative AI models.

[2043] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[2044] Step 1:

[2045] The server receives AI models uploaded from the device. The input is the AI ​​model file, and the output is a database that stores the model and related information (developer name, model name, version, etc.). Specifically, the server assigns a unique ID to the received AI model and performs security checks to ensure that it does not contain any malicious code.

[2046] Step 2:

[2047] The server schedules matches based on a pre-set schedule. The input is schedule information and a list of registered AI models, and the output is the date and time of the next match, the participating AI models, and a record of the task content. The server selects which AI models will participate and records the task content and evaluation criteria in a database.

[2048] Step 3:

[2049] When the match starts, the server notifies the AI ​​models of the set tasks. The input is detailed task information (such as a prompt or the URL of the dataset), and the output is a task notification to each participating AI model. Specifically, the server sends a notification containing task details to each AI model.

[2050] Step 4:

[2051] Each AI model generates an answer to the task and sends it to the server. The input is detailed information about the task, and the output is the generated answer. The AI ​​model analyzes the received task prompt and processes it to generate an answer.

[2052] Step 5:

[2053] The server scores the received answers based on the evaluation criteria. The input is the answer from each AI model and the pre-defined evaluation criteria, and the output is the score of the answer. The server analyzes the answers according to the evaluation criteria and assigns a score to each answer.

[2054] Step 6:

[2055] The server records the match results in a database and publishes them on a web interface. The input is the match score and answer data, and the output is the published match results. The server stores the match results, scores of each AI model, generated answers, evaluation comments, etc. in a database and makes them accessible to users and developers.

[2056] Step 7:

[2057] The server uses an emotion engine to recognize the user's emotions. The input is the user's facial expression, tone of voice, and input data, and the output is the user's emotional state. Specifically, the server analyzes data from the camera and microphone using the emotion engine to identify the user's emotions.

[2058] Step 8:

[2059] The server adjusts the interface based on the user's emotional state recognized by the emotion engine. The input is the recognized emotional state, and the output is an adjusted interface. For example, if the user is happy, the color tone of the interface or the timing of information presentation can be changed.

[2060] Step 9:

[2061] The terminal (for developers) submits its own AI model to the server. The input is the AI ​​model file created by the developer, and the output is the AI ​​model registered on the server. The terminal uploads the AI ​​model and enters related information through a web interface.

[2062] Step 10:

[2063] Users can check the match results and perform detailed analysis. The input is the match result data published on the server, and the output is the information analyzed by the user. Specifically, users can view the match results and evaluation comments using a web interface.

[2064] Step 11:

[2065] The server selects the optimal answer by having the generative AI models compete with each other. The input is the answers from multiple generative AI models, and the output is the selected optimal answer. The server compares the answers from multiple generative AI models and presents the answer with the highest evaluation to the user.

[2066] This trend will enable the competitive environment between generative AI models and interface adjustments based on user emotions.

[2067] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[2068] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[2069] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[2070] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[2071] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on th...

Claims

1. A means for registering an AI model; a means of scheduling matches; A means of notifying participating AI models of challenges; a means for receiving an answer generated by the AI ​​model based on the challenge; means for scoring the generated answers based on evaluation criteria; A system including means for recording and publishing match results.

2. 10. The system of claim 1, further comprising means for verifying game results.

3. The system of claim 1 further comprising means for a user to watch a match and analyze the results.

Images