system

Abstract

The system according to this embodiment aims to appropriately monitor the operation of the AI-Agent and quickly correct any problems. [Solution] The system according to the embodiment comprises a monitoring unit, an evaluation unit, a modification unit, and a provision unit. The monitoring unit monitors the operation of the AI-Agent. The evaluation unit evaluates the problems detected by the monitoring unit. The modification unit modifies the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit. The provision unit provides the modified operation of the AI-Agent to the company.

Description

Technical Field

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

Background Art

[0002] Patent Document 1 discloses a persona chatbot control method performed by at least one processor, including steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to an explanation of a 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

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the prior art, there is a problem that it is difficult to appropriately monitor the operation of an AI-Agent and quickly correct problems.

[0005] The system according to the embodiment aims to appropriately monitor the operation of an AI-Agent and quickly correct problems.

Means for Solving the Problems

[0006] The system according to this embodiment comprises a monitoring unit, an evaluation unit, a modification unit, and a provision unit. The monitoring unit monitors the operation of the AI-Agent. The evaluation unit evaluates the problems detected by the monitoring unit. The modification unit modifies the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit. The provision unit provides the modified operation of the AI-Agent to the company. [Effects of the Invention]

[0007] The system according to this embodiment can appropriately monitor the operation of the AI-Agent and quickly correct any problems. [Brief explanation of the drawing]

[0008] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] This is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] This is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] This shows an emotion map where multiple emotions are mapped. [Figure 10] This shows an emotion map where multiple emotions are mapped. [Modes for carrying out the invention]

[0009] Hereinafter, an example of an embodiment of the system relating to the technology of this disclosure will be described with reference to the attached drawings.

[0010] First, let's explain the terminology used in the following explanation.

[0011] In the following embodiments, the signed processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Furthermore, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include CPU (Central Processing Unit), GPU (Graphics Processing Unit), GPGPU (General-Purpose computing on Graphics Processing Units), APU (Accelerated Processing Unit), or TPU (Tensor Processing Unit).

[0012] In the following embodiments, signed RAM (Random Access Memory) is a memory that temporarily stores information and is used as work memory by the processor.

[0013] In the following embodiments, the signed storage is one or more non-volatile storage devices that store various programs and various parameters. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes.

[0014] In the following embodiments, the labeled communication I / F (Interface) is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between a plurality of computers. Examples of communication standards applied to the communication I / F include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0015] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B". That is, "A and / or B" means that it may be only A, only B, or a combination of A and B. Also, in this specification, when expressing three or more matters connected by "and / or", the same concept as "A and / or B" is applied.

[0016] [First Embodiment] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0017] As shown in FIG. 1, the 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.

[0018] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. 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. Also, the database 24 and the communication I / F 26 are 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).

[0019] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0020] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, and accepts user input. The touch panel 38A accepts user input via touch by detecting contact with an object (e.g., a pen or finger). The microphone 38B accepts user input via voice by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 (see Figure 2) acquires the data indicating the user input.

[0021] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user by outputting the data in a form perceptible to the user (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0022] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0023] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0024] As shown in Figure 2, in the data processing device 12, a specific processing 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" related to the technology of this 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 according to the specific processing program 56 executed on the RAM 30.

[0025] Storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate the user's emotions using the emotion identification model 59 and perform identification processing using the user's emotions. The emotion estimation function (emotion identification function) using the emotion identification model 59 performs various estimations and predictions regarding the user's emotions, including but not limited to these examples. Furthermore, emotion estimation and prediction also include, for example, emotion analysis.

[0026] In the smart device 14, specific processing is performed by the processor 46. The storage 50 stores a specific processing program 60. The specific processing program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the specific processing program 60 from the storage 50 and executes the read specific processing program 60 on the RAM 48. The specific processing is realized by the processor 46 operating as a control unit 46A according to the specific processing program 60 executed on the RAM 48. The smart device 14 also has a data generation model 58 and an emotion identification model 59, similar to the data generation model and emotion identification model 59, and can perform processing similar to that of the specific processing unit 290 using these models.

[0027] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device (e.g., a generation server) may have the data generation model 58. In this case, the data processing device 12 obtains processing results (such as prediction results) using the data generation model 58 by communicating with the server device having the data generation model 58. The data processing device 12 may also be a server device or a terminal device owned by a user (e.g., a mobile phone, robot, home appliance, etc.). Next, an example of processing by the data processing system 10 according to the first embodiment will be described.

[0028] (Example of form 1) The system according to an embodiment of the present invention is a system that monitors AI-Agents by leveraging SoftBank's strength in system monitoring. This system monitors the operation of the AI-Agent and detects problems specific to generative AI, such as hallucination and the number of LLM calls. Next, it attempts to resolve the detected problems using a technique called "reflection," which involves having the LLM analyze its own results. This reflection is performed on the company's AI processing infrastructure. This improves the reliability of the AI-Agent and promotes its adoption in companies. For example, it monitors the operation of the AI-Agent. In this process, it monitors the output generated by the AI-Agent in real time and detects problems such as hallucination and the number of LLM calls. For example, this includes cases where the response generated by the AI-Agent contains inappropriate content or where an excessive number of LLM calls occur. Next, it uses a technique called "reflection," which involves having the LLM analyze its own results, to address the detected problems. Reflection is a technique in which the generative AI evaluates the output results of the generative AI and considers improvements to the model's performance and prompts based on the results. This prevents the occurrence of hallucination and improves the quality of the AI-Agent's output. Furthermore, the AI-Agent's behavior is modified based on the results of the reflection. For example, if the response generated by the AI-Agent is inappropriate, the response is corrected and evaluated again by the generating AI. By repeating this process, the reliability of the AI-Agent can be improved. This mechanism makes it possible to undertake AI-Agent monitoring work and promote the adoption of AI-Agents in companies. For example, it is possible to monitor AI-Agents for various purposes, such as AI-Agents that automatically answer inquiries for companies or AI-Agents that propose travel itineraries. In this way, the system can improve the reliability of AI-Agents and promote their adoption in companies.

[0029] The system according to the embodiment comprises a monitoring unit, an evaluation unit, a correction unit, and a provision unit. The monitoring unit monitors the operation of the AI-Agent. The monitoring unit, for example, monitors the output generated by the AI-Agent in real time. By monitoring the output generated by the AI-Agent in real time, the monitoring unit enables early detection of problems. For example, the monitoring unit detects cases where the response generated by the AI-Agent contains inappropriate content or where an excessive number of LLM calls occur. The evaluation unit evaluates the problems detected by the monitoring unit. For example, the evaluation unit evaluates the output results of the generating AI with the generating AI and considers areas for improvement in the model's performance and prompts based on the results. By evaluating the output results of the generating AI, the evaluation unit can identify areas for improvement in the model's performance and prompts. For example, the evaluation unit evaluates the quality of the responses generated by the generating AI and identifies areas for improvement. The correction unit corrects the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit. For example, the correction unit improves the reliability of the AI-Agent by correcting the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit. The correction unit corrects the response generated by the AI-Agent if it is inappropriate, and then re-evaluates it using the generating AI. The provision unit provides the corrected AI-Agent to the company. The provision unit promotes the adoption of the AI-Agent by providing the corrected AI-Agent to the company. The provision unit can monitor AI-Agents for various purposes, such as an AI-Agent that automatically answers inquiries to companies or an AI-Agent that suggests travel itineraries. As a result, the system according to this embodiment can improve the reliability of the AI-Agent and promote its adoption by companies.

[0030] The monitoring unit monitors the operation of the AI-Agent. For example, the monitoring unit monitors the output generated by the AI-Agent in real time. Specifically, the monitoring unit has a system that continuously collects and analyzes the output logs of the AI-Agent. This allows it to check the content of the text and responses generated by the AI-Agent one by one, and to check for any abnormal or inappropriate content. For example, it can detect cases where the response generated by the AI-Agent contains inappropriate content or where an excessive number of LLM calls are occurring. The monitoring unit uses natural language processing technology to analyze the context and meaning of the generated text and check for the inclusion of specific keywords or phrases. It can also learn the operation patterns of the AI-Agent and detect abnormal patterns that differ from normal operation. For example, it can detect cases where the AI-Agent is processing more requests than usual or where there is an abnormal load during a specific time period. This allows the monitoring unit to monitor the operation of the AI-Agent in real time and enable early detection of problems. Furthermore, the monitoring unit can immediately notify the evaluation unit of any detected problems, prompting a quick response. This allows the monitoring unit to play a crucial role in improving the reliability of the AI-Agent.

[0031] The evaluation unit evaluates the problems detected by the monitoring unit. For example, the evaluation unit evaluates the output results of the generative AI using the generative AI itself, and considers improvements to the model's performance and prompts based on the results. Specifically, the evaluation unit sets evaluation criteria for assessing the quality of the answers generated by the generative AI and evaluates the output results based on those criteria. For example, it evaluates and scores the accuracy, consistency, relevance, and appropriateness of the answers generated by the generative AI. The evaluation unit can also compare the output results of the generative AI with other evaluation methods to evaluate its relative performance. For example, it compares the output results of the generative AI with evaluations by human experts to objectively evaluate the performance of the generative AI. Furthermore, the evaluation unit identifies areas for improvement in the prompts based on the output results of the generative AI. For example, if the generative AI cannot generate consistent answers to a particular question, it analyzes the cause and proposes improvement measures such as modifying the prompts or providing additional training data. In this way, the evaluation unit can identify areas for improvement in the performance of the generative AI and prompts, playing an important role in improving the reliability of the AI-Agent. Furthermore, the evaluation unit provides the evaluation results to the correction unit to assist in correcting the operation of the AI-Agent. This allows the evaluation unit to play a crucial role in improving the reliability of the AI-Agent.

[0032] The modification unit modifies the AI-Agent's operation based on the evaluation results obtained by the evaluation unit. Specifically, the modification unit implements specific methods to modify the AI-Agent's operation based on the problems and areas for improvement identified by the evaluation unit. For example, if the response generated by the AI-Agent is inappropriate, the modification unit modifies the response and re-evaluates it using the generating AI. The modification unit improves the AI-Agent's performance by updating the generating AI's training dataset and retraining it with additional data. It also modifies prompts and adds new prompts to enable the generating AI to generate more appropriate responses. Furthermore, the modification unit adjusts the AI-Agent's operating parameters and tunes them to extract optimal performance. For example, it adjusts the generating AI's temperature parameter and token limits to improve the quality of the generated text. In this way, the modification unit can make specific modifications to improve the AI-Agent's reliability and optimize its operation. Furthermore, the modification unit maintains the AI-Agent's reliability by continuously monitoring the modified AI-Agent's operation and making additional modifications as needed. In this way, the modification unit can play a crucial role in improving the AI-Agent's reliability.

[0033] The Provisioning Department will provide companies with the corrected AI-Agent functionality developed by the Modification Department. Specifically, the Provisioning Department will provide implementation support and assistance to companies to deliver the corrected AI-Agent functionality. For example, they will support the implementation of AI-Agents for various purposes, such as AI-Agents that automatically answer inquiries to companies or AI-Agents that suggest travel itineraries. The Provisioning Department will provide AI-Agents customized according to the needs of companies, helping them to improve operational efficiency and customer satisfaction. Furthermore, the Provisioning Department will provide companies with training on how to use and operate the AI-Agent, enabling companies to effectively utilize it. For example, they will explain how to configure the AI-Agent and troubleshooting procedures, so that company personnel can operate the AI-Agent with confidence. The Provisioning Department will also collect feedback from companies and use it to improve the performance and functionality of the AI-Agent. This will enable the Provisioning Department to provide companies with the corrected AI-Agent functionality and promote its adoption within companies. In addition, the Provisioning Department will provide ongoing support after the AI-Agent is implemented, helping companies to operate the AI-Agent effectively. This allows the service provider to play a crucial role in improving the reliability of AI-Agent and facilitating its adoption by companies.

[0034] The monitoring unit can monitor the output generated by the AI-Agent in real time. For example, the monitoring unit monitors the output generated by the AI-Agent in real time. By monitoring the output generated by the AI-Agent in real time, the monitoring unit can detect problems early. For example, the monitoring unit can detect when the response generated by the AI-Agent contains inappropriate content or when there is an excessive number of LLM calls. This enables early detection of problems by monitoring the output of the AI-Agent in real time. Real-time monitoring includes, but is not limited to, monitoring delay time and update frequency. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without AI. For example, the monitoring unit can monitor the output generated by the AI-Agent in real time and perform monitoring using an AI model that detects problems.

[0035] The evaluation unit can evaluate the output of the generating AI using the generating AI itself, and based on the results, consider areas for improvement in the model's performance and prompts. For example, the evaluation unit can evaluate the output of the generating AI using the generating AI itself, and based on the results, consider areas for improvement in the model's performance and prompts. By evaluating the output of the generating AI, the evaluation unit can identify areas for improvement in the model's performance and prompts. For example, the evaluation unit can evaluate the quality of the answers generated by the generating AI and identify areas for improvement. In this way, by evaluating the output of the generating AI, areas for improvement in the model's performance and prompts can be identified. The generating AI includes, but is not limited to, the model used and the evaluation method used. Some or all of the above-described processes in the evaluation unit may be performed using the generating AI, for example, or without using the generating AI. For example, the evaluation unit can input the output of the generating AI into the generating AI and have the generating AI perform the evaluation.

[0036] The modification unit can modify the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit. For example, the modification unit modifies the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit. The modification unit improves the reliability of the AI-Agent by modifying the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit. For example, if the response generated by the AI-Agent is inappropriate, the modification unit modifies the response and performs the evaluation again using the generating AI. This improves the reliability of the AI-Agent by modifying the operation of the AI-Agent based on the evaluation results. The evaluation results include, but are not limited to, evaluation scores and evaluation indicators. Some or all of the above processing in the modification unit may be performed using AI, for example, or without using AI. For example, the modification unit can input the evaluation results obtained by the evaluation unit into the generating AI and have the generating AI perform the modification.

[0037] The provisioning unit can provide the modified AI-Agent's behavior to companies. For example, the provisioning unit can provide the modified AI-Agent's behavior to companies. By providing the modified AI-Agent's behavior to companies, the provisioning unit facilitates the adoption of AI-Agents in companies. The provisioning unit can monitor AI-Agents for various purposes, such as AI-Agents that automatically answer inquiries to companies or AI-Agents that suggest travel itineraries. This facilitates the adoption of AI-Agents in companies by providing them with the modified AI-Agent's behavior. Companies include, but are not limited to, company size and industry. Some or all of the above-described processes in the provisioning unit may be performed using AI, for example, or not using AI. For example, the provisioning unit can input the modified AI-Agent's behavior into a generating AI and have the generating AI execute the provision.

[0038] The monitoring unit can detect problems specific to generative AI, such as hallucination and the number of LLM calls. The monitoring unit can detect problems specific to generative AI, such as hallucination and the number of LLM calls. The monitoring unit improves the reliability of the AI-Agent by detecting problems specific to generative AI. The monitoring unit can detect, for example, cases where the responses generated by the AI-Agent contain inappropriate content or where there is an excessive number of LLM calls. This improves the reliability of the AI-Agent by detecting problems specific to generative AI. Hallucination includes, but is not limited to, examples such as criteria for misgeneration and detection algorithms. The number of LLM calls includes, but is not limited to, examples such as methods for measuring the number of calls and thresholds. Some or all of the above processing in the monitoring unit may be performed using, for example, generative AI, or without using generative AI. For example, the monitoring unit can perform monitoring using a generative AI model that detects problems specific to generative AI.

[0039] The monitoring unit can analyze the AI-Agent's operational history and change its monitoring focus based on specific patterns. For example, the monitoring unit can prioritize monitoring problems that have occurred frequently in the past. The monitoring unit can prioritize monitoring problems that are likely to occur during specific time periods. The monitoring unit can prioritize monitoring problems related to specific operations or events. This enables focused monitoring based on specific patterns by analyzing the operational history. The operational history includes, but is not limited to, the format of log data and the analysis algorithm. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or not using AI. For example, the monitoring unit can input the AI-Agent's operational history into a generating AI and have the generating AI perform pattern analysis.

[0040] The monitoring unit can improve the accuracy of monitoring by considering the operating environment of the AI-Agent during monitoring. For example, if the network condition is unstable, the monitoring unit can increase the frequency to improve monitoring accuracy. If the server load is high, the monitoring unit can improve monitoring accuracy by focusing on important monitoring items. The monitoring unit can dynamically adjust the accuracy of monitoring in response to changes in the operating environment. This improves monitoring accuracy by considering the operating environment. The operating environment includes, but is not limited to, hardware configuration and network environment. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without AI. For example, the monitoring unit can input AI-Agent operating environment data into a generating AI and have the generating AI perform the improvement of monitoring accuracy.

[0041] The monitoring unit can improve the accuracy of monitoring by referring to external data related to the operation of the AI-Agent during monitoring. For example, the monitoring unit can refer to weather information and improve the accuracy of monitoring by considering the impact of weather. The monitoring unit can refer to market trends and improve the accuracy of monitoring according to economic conditions. The monitoring unit can acquire external data in real time and dynamically adjust the accuracy of monitoring. As a result, the accuracy of monitoring is improved by referring to external data. External data includes, but is not limited to, data sources and data formats. Some or all of the above processing in the monitoring unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the monitoring unit can input external data into a generative AI and have the generative AI perform the improvement of monitoring accuracy.

[0042] The monitoring unit can collect user feedback regarding the operation of the AI-Agent in real time during monitoring and reflect it in the monitoring results. For example, the monitoring unit can collect user feedback in real time and reflect it in the monitoring results. The monitoring unit can analyze user feedback and adjust the priority of monitoring items. The monitoring unit can improve the accuracy of monitoring based on user feedback. This means that by collecting user feedback in real time, it can be reflected in the monitoring results. User feedback includes, but is not limited to, the format of the feedback and the collection tools. Some or all of the above processing in the monitoring unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the monitoring unit can input user feedback into a generative AI and have the generative AI perform the action of reflecting it in the monitoring results.

[0043] The evaluation unit can optimize the evaluation algorithm by referring to past evaluation results of the AI-Agent during the evaluation process. For example, the evaluation unit optimizes the evaluation algorithm based on past evaluation results. The evaluation unit can analyze past evaluation results and adjust the evaluation criteria. The evaluation unit can improve the accuracy of the evaluation by referring to past evaluation results. This allows the evaluation algorithm to be optimized by referring to past evaluation results. The evaluation algorithm includes, but is not limited to, the algorithms used and optimization methods. Some or all of the above-described processes in the evaluation unit may be performed using, for example, a generative AI, or without a generative AI. For example, the evaluation unit can input past evaluation results into a generative AI and have the generative AI perform the optimization of the evaluation algorithm.

[0044] The evaluation unit can improve the accuracy of its evaluation by referring to external data regarding the operation of the AI-Agent during the evaluation process. For example, the evaluation unit can improve the accuracy of its evaluation by referring to the performance of competitors' AI-Agents. The evaluation unit can adjust the evaluation criteria based on the external data. The evaluation unit can acquire external data in real time and dynamically adjust the accuracy of its evaluation. This improves the accuracy of the evaluation by referring to external data. External data includes, but is not limited to, data sources and data formats. Some or all of the above processing in the evaluation unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the evaluation unit can input external data into a generative AI and have the generative AI perform the task of improving the accuracy of the evaluation.

[0045] The evaluation unit can collect user feedback regarding the AI-Agent's operation during evaluation and reflect it in the evaluation results. The evaluation unit can, for example, collect user feedback and reflect it in the evaluation results. The evaluation unit can analyze user feedback and adjust the evaluation criteria. The evaluation unit can improve the accuracy of the evaluation based on user feedback. Thus, by collecting user feedback, it can be reflected in the evaluation results. User feedback includes, but is not limited to, the format of the feedback and the collection tools. Some or all of the above processing in the evaluation unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the evaluation unit can input user feedback into a generative AI and have the generative AI perform the reflection of it in the evaluation results.

[0046] The evaluation unit can improve the accuracy of its evaluation by referring to relevant literature on the operation of the AI-Agent during the evaluation process. For example, the evaluation unit can improve the accuracy of its evaluation by referring to relevant literature. The evaluation unit can adjust the evaluation criteria based on the relevant literature. The evaluation unit can acquire relevant literature in real time and dynamically adjust the accuracy of its evaluation. As a result, the accuracy of the evaluation is improved by referring to relevant literature. Relevant literature includes, but is not limited to, literature databases and referencing methods. Some or all of the above processing in the evaluation unit may be performed using, for example, a generative AI, or without a generative AI. For example, the evaluation unit can input relevant literature into a generative AI and have the generative AI perform the improvement of the evaluation accuracy.

[0047] The correction unit can select the optimal correction method by referring to the AI-Agent's past correction history during the correction process. For example, the correction unit selects the optimal correction method based on past correction history. The correction unit can analyze past correction history and adjust the correction method. The correction unit can improve the accuracy of corrections by referring to past correction history. This allows the optimal correction method to be selected by referring to past correction history. The correction history includes, but is not limited to, the format of the history and the method of reference. Some or all of the above-described processes in the correction unit may be performed using AI, for example, or without using AI. For example, the correction unit can input past correction history into a generating AI and have the generating AI select the optimal correction method.

[0048] The correction unit can improve the accuracy of corrections by considering the operating environment of the AI-Agent during the correction process. For example, if the network conditions are unstable, the correction unit can increase the frequency of corrections to improve accuracy. If the server load is high, the correction unit can focus on important correction items to improve accuracy. The correction unit can dynamically adjust the accuracy of corrections in response to changes in the operating environment. This improves the accuracy of corrections by considering the operating environment. The operating environment includes, but is not limited to, hardware configuration and network environment. Some or all of the above-described processes in the correction unit may be performed using AI, for example, or without AI. For example, the correction unit can input AI-Agent operating environment data into a generating AI and have the generating AI perform the correction accuracy improvement.

[0049] The modification unit can collect user feedback regarding the AI-Agent's operation during the modification process and incorporate it into the modification. For example, the modification unit can collect user feedback and incorporate it into the modification. The modification unit can analyze user feedback and adjust the modification method. The modification unit can improve the accuracy of the modification based on user feedback. Thus, by collecting user feedback, it can be incorporated into the modification. User feedback includes, but is not limited to, the format of the feedback and the tools used to collect it. Some or all of the above-described processes in the modification unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the modification unit can input user feedback into a generative AI and have the generative AI incorporate it into the modification.

[0050] The correction unit can improve the accuracy of corrections by referring to relevant literature on the operation of the AI-Agent during the correction process. The correction unit can improve the accuracy of corrections by referring to relevant literature, for example. The correction unit can adjust the correction method based on the relevant literature. The correction unit can acquire relevant literature in real time and dynamically adjust the accuracy of corrections. As a result, the accuracy of corrections is improved by referring to relevant literature. Relevant literature includes, but is not limited to, literature databases and referencing methods. Some or all of the above processing in the correction unit may be performed using, for example, a generative AI, or without a generative AI. For example, the correction unit can input relevant literature into a generative AI and have the generative AI perform the correction accuracy improvement.

[0051] The delivery unit can select the optimal delivery method by referring to the AI-Agent's past delivery history at the time of delivery. The delivery unit can, for example, select the optimal delivery method based on past delivery history. The delivery unit can analyze past delivery history and adjust the delivery method. The delivery unit can improve the accuracy of delivery by referring to past delivery history. This allows the optimal delivery method to be selected by referring to past delivery history. The delivery history includes, but is not limited to, the format of the history and the method of referencing. Some or all of the above processing in the delivery unit may be performed using AI, for example, or without using AI. For example, the delivery unit can input past delivery history into a generating AI and have the generating AI select the optimal delivery method.

[0052] The delivery unit can improve the accuracy of deliveries by considering the operating environment of the AI-Agent at the time of delivery. For example, if the network conditions are unstable, the delivery unit can increase the frequency of deliveries to improve accuracy. If the server load is high, the delivery unit can improve accuracy by focusing on important delivery items. The delivery unit can dynamically adjust the accuracy of deliveries in response to changes in the operating environment. This improves the accuracy of deliveries by considering the operating environment. The operating environment includes, but is not limited to, hardware configuration and network environment. Some or all of the above processing in the delivery unit may be performed using AI, for example, or without AI. For example, the delivery unit can input AI-Agent operating environment data into a generating AI and have the generating AI perform the task of improving the accuracy of deliveries.

[0053] The service provider can collect user feedback regarding the operation of the AI-Agent at the time of delivery and reflect it in the delivered content. The service provider can, for example, collect user feedback and reflect it in the delivered content. The service provider can analyze user feedback and adjust the delivery method. The service provider can improve the accuracy of the delivery based on user feedback. This allows the service provider to reflect user feedback by collecting it. User feedback includes, but is not limited to, the format of the feedback and the tools used to collect it. Some or all of the above processing in the service provider may be performed using, for example, a generative AI, or not using a generative AI. For example, the service provider can input user feedback into a generative AI and have the generative AI perform the action of reflecting it in the delivered content.

[0054] The delivery unit can improve the accuracy of its delivery by referring to relevant literature on the operation of the AI-Agent at the time of delivery. The delivery unit can improve the accuracy of its delivery by referring to relevant literature. The delivery unit can adjust the delivery method based on the relevant literature. The delivery unit can acquire relevant literature in real time and dynamically adjust the accuracy of its delivery. As a result, the accuracy of its delivery is improved by referring to relevant literature. Relevant literature includes, but is not limited to, literature databases and referencing methods. Some or all of the above processing in the delivery unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the delivery unit can input relevant literature into a generative AI and have the generative AI perform the improvement of the accuracy of its delivery.

[0055] The system according to the embodiment is not limited to the example described above, and various modifications are possible, for example, as follows.

[0056] The monitoring unit not only monitors the operation of the AI-Agent, but can also collect user feedback on the AI-Agent's operation in real time and reflect it in the monitoring results. For example, if a user is dissatisfied with the AI-Agent's operation, they can immediately send that feedback to the monitoring unit, which can then adjust its monitoring focus based on that feedback. Conversely, if a user is satisfied with the AI-Agent's operation, the monitoring frequency can be reduced based on that feedback, saving system resources. Furthermore, by analyzing user feedback, areas for improvement in the AI-Agent's operation can be identified, thereby improving the accuracy of monitoring. In this way, by collecting user feedback in real time and reflecting it in the monitoring results, more appropriate monitoring becomes possible.

[0057] The evaluation unit not only evaluates the output results of the generated AI, but can also improve the accuracy of the evaluation by referring to external data related to the operation of the AI-Agent. For example, it can improve the accuracy of the evaluation by referring to the performance of competitors' AI-Agents. It can also adjust the evaluation criteria based on the external data. Furthermore, it can acquire external data in real time and dynamically adjust the accuracy of the evaluation. As a result, the accuracy of the evaluation is improved by referring to external data. External data includes, but is not limited to, data sources and data formats.

[0058] The modification unit not only corrects the AI-Agent's operation based on the evaluation results obtained by the evaluation unit, but can also improve the accuracy of the modifications by considering the AI-Agent's operating environment. For example, if the network condition is unstable, the frequency of modifications can be increased to improve accuracy. Also, if the server load is high, the accuracy of modifications can be improved by focusing on important items. Furthermore, the accuracy of modifications can be dynamically adjusted in response to changes in the operating environment. In this way, the accuracy of modifications is improved by considering the operating environment.

[0059] The service provider can not only provide the corrected AI-Agent functionality to companies, but also collect user feedback on the AI-Agent's operation and incorporate it into the service. For example, it can collect user feedback and incorporate it into the service. It can also analyze user feedback and adjust the service delivery method. Furthermore, it can improve the accuracy of the service based on user feedback. In this way, by collecting user feedback, it can be incorporated into the service delivery.

[0060] The monitoring unit not only monitors the operation of the AI-Agent, but can also improve the accuracy of monitoring by referring to relevant literature on the AI-Agent's operation. For example, it can improve monitoring accuracy by referring to relevant literature. It can also adjust the monitoring method based on the relevant literature. Furthermore, it can acquire relevant literature in real time and dynamically adjust the accuracy of monitoring. As a result, monitoring accuracy is improved by referring to relevant literature.

[0061] The following briefly describes the processing flow for example form 1.

[0062] Step 1: The monitoring unit monitors the operation of the AI-Agent. For example, the monitoring unit monitors the output generated by the AI-Agent in real time, enabling early detection of problems. Specifically, it detects cases such as when the response generated by the AI-Agent contains inappropriate content or when an excessive number of LLM calls occur. Step 2: The evaluation unit evaluates the problems detected by the monitoring unit. For example, the evaluation unit evaluates the output results of the generation AI using the generation AI itself, and considers improvements to the model's performance and prompts based on the results. This allows the quality of the answers generated by the generation AI to be evaluated and areas for improvement to be identified. Step 3: The modification unit modifies the AI-Agent's operation based on the evaluation results obtained by the evaluation unit. For example, the modification unit improves the reliability of the AI-Agent by modifying its operation based on the evaluation results obtained by the evaluation unit. Specifically, if the response generated by the AI-Agent is inappropriate, the modification unit modifies the response and performs the evaluation again using the generating AI. Step 4: The provisioning unit provides the company with the corrected AI-Agent functionality. The provisioning unit facilitates the adoption of the AI-Agent by providing the company with the corrected AI-Agent functionality. Specifically, it is possible to monitor AI-Agents for various purposes, such as AI-Agents that automatically answer inquiries for companies or AI-Agents that suggest travel itineraries.

[0063] (Example of form 2) The system according to an embodiment of the present invention is a system that monitors AI-Agents by leveraging SoftBank's strength in system monitoring. This system monitors the operation of the AI-Agent and detects problems specific to generative AI, such as hallucination and the number of LLM calls. Next, it attempts to resolve the detected problems using a technique called "reflection," which involves having the LLM analyze its own results. This reflection is performed on the company's AI processing infrastructure. This improves the reliability of the AI-Agent and promotes its adoption in companies. For example, it monitors the operation of the AI-Agent. In this process, it monitors the output generated by the AI-Agent in real time and detects problems such as hallucination and the number of LLM calls. For example, this includes cases where the response generated by the AI-Agent contains inappropriate content or where an excessive number of LLM calls occur. Next, it uses a technique called "reflection," which involves having the LLM analyze its own results, to address the detected problems. Reflection is a technique in which the generative AI evaluates the output results of the generative AI and considers improvements to the model's performance and prompts based on the results. This prevents the occurrence of hallucination and improves the quality of the AI-Agent's output. Furthermore, the AI-Agent's behavior is modified based on the results of the reflection. For example, if the response generated by the AI-Agent is inappropriate, the response is corrected and evaluated again by the generating AI. By repeating this process, the reliability of the AI-Agent can be improved. This mechanism makes it possible to undertake AI-Agent monitoring work and promote the adoption of AI-Agents in companies. For example, it is possible to monitor AI-Agents for various purposes, such as AI-Agents that automatically answer inquiries for companies or AI-Agents that propose travel itineraries. In this way, the system can improve the reliability of AI-Agents and promote their adoption in companies.

[0064] The system according to the embodiment comprises a monitoring unit, an evaluation unit, a correction unit, and a provision unit. The monitoring unit monitors the operation of the AI-Agent. The monitoring unit, for example, monitors the output generated by the AI-Agent in real time. By monitoring the output generated by the AI-Agent in real time, the monitoring unit enables early detection of problems. For example, the monitoring unit detects cases where the response generated by the AI-Agent contains inappropriate content or where an excessive number of LLM calls occur. The evaluation unit evaluates the problems detected by the monitoring unit. For example, the evaluation unit evaluates the output results of the generating AI with the generating AI and considers areas for improvement in the model's performance and prompts based on the results. By evaluating the output results of the generating AI, the evaluation unit can identify areas for improvement in the model's performance and prompts. For example, the evaluation unit evaluates the quality of the responses generated by the generating AI and identifies areas for improvement. The correction unit corrects the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit. For example, the correction unit improves the reliability of the AI-Agent by correcting the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit. The correction unit corrects the response generated by the AI-Agent if it is inappropriate, and then re-evaluates it using the generating AI. The provision unit provides the corrected AI-Agent to the company. The provision unit promotes the adoption of the AI-Agent by providing the corrected AI-Agent to the company. The provision unit can monitor AI-Agents for various purposes, such as an AI-Agent that automatically answers inquiries to companies or an AI-Agent that suggests travel itineraries. As a result, the system according to this embodiment can improve the reliability of the AI-Agent and promote its adoption by companies.

[0065] The monitoring unit monitors the operation of the AI-Agent. For example, the monitoring unit monitors the output generated by the AI-Agent in real time. Specifically, the monitoring unit has a system that continuously collects and analyzes the output logs of the AI-Agent. This allows it to check the content of the text and responses generated by the AI-Agent one by one, and to check for any abnormal or inappropriate content. For example, it can detect cases where the response generated by the AI-Agent contains inappropriate content or where an excessive number of LLM calls are occurring. The monitoring unit uses natural language processing technology to analyze the context and meaning of the generated text and check for the inclusion of specific keywords or phrases. It can also learn the operation patterns of the AI-Agent and detect abnormal patterns that differ from normal operation. For example, it can detect cases where the AI-Agent is processing more requests than usual or where there is an abnormal load during a specific time period. This allows the monitoring unit to monitor the operation of the AI-Agent in real time and enable early detection of problems. Furthermore, the monitoring unit can immediately notify the evaluation unit of any detected problems, prompting a quick response. This allows the monitoring unit to play a crucial role in improving the reliability of the AI-Agent.

[0066] The evaluation unit evaluates the problems detected by the monitoring unit. For example, the evaluation unit evaluates the output results of the generative AI using the generative AI itself, and considers improvements to the model's performance and prompts based on the results. Specifically, the evaluation unit sets evaluation criteria for assessing the quality of the answers generated by the generative AI and evaluates the output results based on those criteria. For example, it evaluates and scores the accuracy, consistency, relevance, and appropriateness of the answers generated by the generative AI. The evaluation unit can also compare the output results of the generative AI with other evaluation methods to evaluate its relative performance. For example, it compares the output results of the generative AI with evaluations by human experts to objectively evaluate the performance of the generative AI. Furthermore, the evaluation unit identifies areas for improvement in the prompts based on the output results of the generative AI. For example, if the generative AI cannot generate consistent answers to a particular question, it analyzes the cause and proposes improvement measures such as modifying the prompts or providing additional training data. In this way, the evaluation unit can identify areas for improvement in the performance of the generative AI and prompts, playing an important role in improving the reliability of the AI-Agent. Furthermore, the evaluation unit provides the evaluation results to the correction unit to assist in correcting the operation of the AI-Agent. This allows the evaluation unit to play a crucial role in improving the reliability of the AI-Agent.

[0067] The modification unit modifies the AI-Agent's operation based on the evaluation results obtained by the evaluation unit. Specifically, the modification unit implements specific methods to modify the AI-Agent's operation based on the problems and areas for improvement identified by the evaluation unit. For example, if the response generated by the AI-Agent is inappropriate, the modification unit modifies the response and re-evaluates it using the generating AI. The modification unit improves the AI-Agent's performance by updating the generating AI's training dataset and retraining it with additional data. It also modifies prompts and adds new prompts to enable the generating AI to generate more appropriate responses. Furthermore, the modification unit adjusts the AI-Agent's operating parameters and tunes them to extract optimal performance. For example, it adjusts the generating AI's temperature parameter and token limits to improve the quality of the generated text. In this way, the modification unit can make specific modifications to improve the AI-Agent's reliability and optimize its operation. Furthermore, the modification unit maintains the AI-Agent's reliability by continuously monitoring the modified AI-Agent's operation and making additional modifications as needed. In this way, the modification unit can play a crucial role in improving the AI-Agent's reliability.

[0068] The Provisioning Department will provide companies with the corrected AI-Agent functionality developed by the Modification Department. Specifically, the Provisioning Department will provide implementation support and assistance to companies to deliver the corrected AI-Agent functionality. For example, they will support the implementation of AI-Agents for various purposes, such as AI-Agents that automatically answer inquiries to companies or AI-Agents that suggest travel itineraries. The Provisioning Department will provide AI-Agents customized according to the needs of companies, helping them to improve operational efficiency and customer satisfaction. Furthermore, the Provisioning Department will provide companies with training on how to use and operate the AI-Agent, enabling companies to effectively utilize it. For example, they will explain how to configure the AI-Agent and troubleshooting procedures, so that company personnel can operate the AI-Agent with confidence. The Provisioning Department will also collect feedback from companies and use it to improve the performance and functionality of the AI-Agent. This will enable the Provisioning Department to provide companies with the corrected AI-Agent functionality and promote its adoption within companies. In addition, the Provisioning Department will provide ongoing support after the AI-Agent is implemented, helping companies to operate the AI-Agent effectively. This allows the service provider to play a crucial role in improving the reliability of AI-Agent and facilitating its adoption by companies.

[0069] The monitoring unit can monitor the output generated by the AI-Agent in real time. For example, the monitoring unit monitors the output generated by the AI-Agent in real time. By monitoring the output generated by the AI-Agent in real time, the monitoring unit can detect problems early. For example, the monitoring unit can detect when the response generated by the AI-Agent contains inappropriate content or when there is an excessive number of LLM calls. This enables early detection of problems by monitoring the output of the AI-Agent in real time. Real-time monitoring includes, but is not limited to, monitoring delay time and update frequency. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without AI. For example, the monitoring unit can monitor the output generated by the AI-Agent in real time and perform monitoring using an AI model that detects problems.

[0070] The evaluation unit can evaluate the output of the generating AI using the generating AI itself, and based on the results, consider areas for improvement in the model's performance and prompts. For example, the evaluation unit can evaluate the output of the generating AI using the generating AI itself, and based on the results, consider areas for improvement in the model's performance and prompts. By evaluating the output of the generating AI, the evaluation unit can identify areas for improvement in the model's performance and prompts. For example, the evaluation unit can evaluate the quality of the answers generated by the generating AI and identify areas for improvement. In this way, by evaluating the output of the generating AI, areas for improvement in the model's performance and prompts can be identified. The generating AI includes, but is not limited to, the model used and the evaluation method used. Some or all of the above-described processes in the evaluation unit may be performed using the generating AI, for example, or without using the generating AI. For example, the evaluation unit can input the output of the generating AI into the generating AI and have the generating AI perform the evaluation.

[0071] The modification unit can modify the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit. For example, the modification unit modifies the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit. The modification unit improves the reliability of the AI-Agent by modifying the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit. For example, if the response generated by the AI-Agent is inappropriate, the modification unit modifies the response and performs the evaluation again using the generating AI. This improves the reliability of the AI-Agent by modifying the operation of the AI-Agent based on the evaluation results. The evaluation results include, but are not limited to, evaluation scores and evaluation indicators. Some or all of the above processing in the modification unit may be performed using AI, for example, or without using AI. For example, the modification unit can input the evaluation results obtained by the evaluation unit into the generating AI and have the generating AI perform the modification.

[0072] The provisioning unit can provide the modified AI-Agent's behavior to companies. For example, the provisioning unit can provide the modified AI-Agent's behavior to companies. By providing the modified AI-Agent's behavior to companies, the provisioning unit facilitates the adoption of AI-Agents in companies. The provisioning unit can monitor AI-Agents for various purposes, such as AI-Agents that automatically answer inquiries to companies or AI-Agents that suggest travel itineraries. This facilitates the adoption of AI-Agents in companies by providing them with the modified AI-Agent's behavior. Companies include, but are not limited to, company size and industry. Some or all of the above-described processes in the provisioning unit may be performed using AI, for example, or not using AI. For example, the provisioning unit can input the modified AI-Agent's behavior into a generating AI and have the generating AI execute the provision.

[0073] The monitoring unit can detect problems specific to generative AI, such as hallucination and the number of LLM calls. The monitoring unit can detect problems specific to generative AI, such as hallucination and the number of LLM calls. The monitoring unit improves the reliability of the AI-Agent by detecting problems specific to generative AI. The monitoring unit can detect, for example, cases where the responses generated by the AI-Agent contain inappropriate content or where there is an excessive number of LLM calls. This improves the reliability of the AI-Agent by detecting problems specific to generative AI. Hallucination includes, but is not limited to, examples such as criteria for misgeneration and detection algorithms. The number of LLM calls includes, but is not limited to, examples such as methods for measuring the number of calls and thresholds. Some or all of the above processing in the monitoring unit may be performed using, for example, generative AI, or without using generative AI. For example, the monitoring unit can perform monitoring using a generative AI model that detects problems specific to generative AI.

[0074] The monitoring unit can estimate the user's emotions and adjust the monitoring frequency based on the estimated emotions. For example, if the user is stressed, the monitoring unit can increase the monitoring frequency to detect problems early. If the user is relaxed, the monitoring unit can decrease the monitoring frequency to conserve system resources. If the user is in a hurry, the monitoring unit can adjust the monitoring frequency to focus on important monitoring items. This allows for more appropriate monitoring by adjusting the monitoring frequency based on the user's emotions. User emotions include, but are not limited to, emotion analysis algorithms and emotion classification criteria. Emotion estimation is achieved using an emotion estimation function, for example, an emotion engine or generative AI. Generative AI includes, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without AI. For example, the monitoring unit can input user emotion data into a generative AI and have the generative AI perform emotion estimation.

[0075] The monitoring unit can analyze the AI-Agent's operational history and change its monitoring focus based on specific patterns. For example, the monitoring unit can prioritize monitoring problems that have occurred frequently in the past. The monitoring unit can prioritize monitoring problems that are likely to occur during specific time periods. The monitoring unit can prioritize monitoring problems related to specific operations or events. This enables focused monitoring based on specific patterns by analyzing the operational history. The operational history includes, but is not limited to, the format of log data and the analysis algorithm. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or not using AI. For example, the monitoring unit can input the AI-Agent's operational history into a generating AI and have the generating AI perform pattern analysis.

[0076] The monitoring unit can improve the accuracy of monitoring by considering the operating environment of the AI-Agent during monitoring. For example, if the network condition is unstable, the monitoring unit can increase the frequency to improve monitoring accuracy. If the server load is high, the monitoring unit can improve monitoring accuracy by focusing on important monitoring items. The monitoring unit can dynamically adjust the accuracy of monitoring in response to changes in the operating environment. This improves monitoring accuracy by considering the operating environment. The operating environment includes, but is not limited to, hardware configuration and network environment. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without AI. For example, the monitoring unit can input AI-Agent operating environment data into a generating AI and have the generating AI perform the improvement of monitoring accuracy.

[0077] The monitoring unit can estimate the user's emotions and determine the priority of items to monitor based on the estimated user emotions. For example, if the user is feeling anxious, the monitoring unit will prioritize monitoring important items. If the user is feeling at ease, the monitoring unit can prioritize monitoring normal items. If the user is in a hurry, the monitoring unit can prioritize monitoring items of high urgency. This allows for more appropriate monitoring by determining the priority of monitoring items based on the user's emotions. Prioritization of monitoring items includes, but is not limited to, criteria for evaluating importance and methods for prioritizing. Emotion estimation is achieved using an emotion estimation function, for example, an emotion engine or generative AI. Generative AI includes, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the monitoring unit may be performed using AI, or not using AI. For example, the monitoring unit can input user emotion data into a generative AI and have the generative AI perform emotion estimation.

[0078] The monitoring unit can improve the accuracy of monitoring by referring to external data related to the operation of the AI-Agent during monitoring. For example, the monitoring unit can refer to weather information and improve the accuracy of monitoring by considering the impact of weather. The monitoring unit can refer to market trends and improve the accuracy of monitoring according to economic conditions. The monitoring unit can acquire external data in real time and dynamically adjust the accuracy of monitoring. As a result, the accuracy of monitoring is improved by referring to external data. External data includes, but is not limited to, data sources and data formats. Some or all of the above processing in the monitoring unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the monitoring unit can input external data into a generative AI and have the generative AI perform the improvement of monitoring accuracy.

[0079] The monitoring unit can collect user feedback regarding the operation of the AI-Agent in real time during monitoring and reflect it in the monitoring results. For example, the monitoring unit can collect user feedback in real time and reflect it in the monitoring results. The monitoring unit can analyze user feedback and adjust the priority of monitoring items. The monitoring unit can improve the accuracy of monitoring based on user feedback. This means that by collecting user feedback in real time, it can be reflected in the monitoring results. User feedback includes, but is not limited to, the format of the feedback and the collection tools. Some or all of the above processing in the monitoring unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the monitoring unit can input user feedback into a generative AI and have the generative AI perform the action of reflecting it in the monitoring results.

[0080] The evaluation unit can estimate the user's emotions and adjust the evaluation criteria based on the estimated emotions. For example, if the user is feeling anxious, the evaluation unit can apply strict evaluation criteria. If the user is feeling at ease, the evaluation unit can apply normal evaluation criteria. If the user is in a hurry, the evaluation unit can apply rapid evaluation criteria. This allows for more appropriate evaluations by adjusting the evaluation criteria based on the user's emotions. Evaluation criteria include, but are not limited to, evaluation metrics and procedures for changing criteria. Emotion estimation is achieved using an emotion estimation function, for example, an emotion engine or generative AI. Generative AI includes, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the evaluation unit may be performed using, for example, generative AI, or not using generative AI. For example, the evaluation unit can input user emotion data into a generative AI and have the generative AI perform emotion estimation.

[0081] The evaluation unit can optimize the evaluation algorithm by referring to past evaluation results of the AI-Agent during the evaluation process. For example, the evaluation unit optimizes the evaluation algorithm based on past evaluation results. The evaluation unit can analyze past evaluation results and adjust the evaluation criteria. The evaluation unit can improve the accuracy of the evaluation by referring to past evaluation results. This allows the evaluation algorithm to be optimized by referring to past evaluation results. The evaluation algorithm includes, but is not limited to, the algorithms used and optimization methods. Some or all of the above-described processes in the evaluation unit may be performed using, for example, a generative AI, or without a generative AI. For example, the evaluation unit can input past evaluation results into a generative AI and have the generative AI perform the optimization of the evaluation algorithm.

[0082] The evaluation unit can improve the accuracy of its evaluation by referring to external data regarding the operation of the AI-Agent during the evaluation process. For example, the evaluation unit can improve the accuracy of its evaluation by referring to the performance of competitors' AI-Agents. The evaluation unit can adjust the evaluation criteria based on the external data. The evaluation unit can acquire external data in real time and dynamically adjust the accuracy of its evaluation. This improves the accuracy of the evaluation by referring to external data. External data includes, but is not limited to, data sources and data formats. Some or all of the above processing in the evaluation unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the evaluation unit can input external data into a generative AI and have the generative AI perform the task of improving the accuracy of the evaluation.

[0083] The evaluation unit can estimate the user's emotions and adjust the display method of the evaluation results based on the estimated user emotions. For example, if the user is feeling anxious, the evaluation unit can display detailed evaluation results. If the user is feeling at ease, the evaluation unit can display concise evaluation results. If the user is in a hurry, the evaluation unit can display concise evaluation results. By adjusting the display method of the evaluation results based on the user's emotions, a more appropriate display becomes possible. The display method of the evaluation results includes, but is not limited to, the display format and timing of display. Emotion estimation is achieved using an emotion estimation function, for example, an emotion engine or a generative AI. The generative AI is a text generation AI (e.g., LLM) or a multimodal generation AI, but is not limited to such examples. Some or all of the above processing in the evaluation unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the evaluation unit can input user emotion data into a generative AI and have the generative AI perform emotion estimation.

[0084] The evaluation unit can collect user feedback regarding the AI-Agent's operation during evaluation and reflect it in the evaluation results. The evaluation unit can, for example, collect user feedback and reflect it in the evaluation results. The evaluation unit can analyze user feedback and adjust the evaluation criteria. The evaluation unit can improve the accuracy of the evaluation based on user feedback. Thus, by collecting user feedback, it can be reflected in the evaluation results. User feedback includes, but is not limited to, the format of the feedback and the collection tools. Some or all of the above processing in the evaluation unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the evaluation unit can input user feedback into a generative AI and have the generative AI perform the reflection of it in the evaluation results.

[0085] The evaluation unit can improve the accuracy of its evaluation by referring to relevant literature on the operation of the AI-Agent during the evaluation process. For example, the evaluation unit can improve the accuracy of its evaluation by referring to relevant literature. The evaluation unit can adjust the evaluation criteria based on the relevant literature. The evaluation unit can acquire relevant literature in real time and dynamically adjust the accuracy of its evaluation. As a result, the accuracy of the evaluation is improved by referring to relevant literature. Relevant literature includes, but is not limited to, literature databases and referencing methods. Some or all of the above processing in the evaluation unit may be performed using, for example, a generative AI, or without a generative AI. For example, the evaluation unit can input relevant literature into a generative AI and have the generative AI perform the improvement of the evaluation accuracy.

[0086] The editing unit can estimate the user's emotions and determine the priority of corrections based on the estimated emotions. For example, if the user is feeling anxious, the editing unit may prioritize important corrections. If the user is feeling at ease, the editing unit may prioritize normal corrections. If the user is in a hurry, the editing unit may prioritize urgent corrections. This allows for more appropriate corrections by determining the priority of corrections based on the user's emotions. Correction prioritization includes, but is not limited to, criteria for evaluating importance and methods for prioritizing. Emotion estimation is achieved using an emotion estimation function, for example, an emotion engine or generative AI. Generative AI includes, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the editing unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the editing unit can input user emotion data into a generative AI and have the generative AI perform emotion estimation.

[0087] The correction unit can select the optimal correction method by referring to the AI-Agent's past correction history during the correction process. For example, the correction unit selects the optimal correction method based on past correction history. The correction unit can analyze past correction history and adjust the correction method. The correction unit can improve the accuracy of corrections by referring to past correction history. This allows the optimal correction method to be selected by referring to past correction history. The correction history includes, but is not limited to, the format of the history and the method of reference. Some or all of the above-described processes in the correction unit may be performed using AI, for example, or without using AI. For example, the correction unit can input past correction history into a generating AI and have the generating AI select the optimal correction method.

[0088] The correction unit can improve the accuracy of corrections by considering the operating environment of the AI-Agent during the correction process. For example, if the network conditions are unstable, the correction unit can increase the frequency of corrections to improve accuracy. If the server load is high, the correction unit can focus on important correction items to improve accuracy. The correction unit can dynamically adjust the accuracy of corrections in response to changes in the operating environment. This improves the accuracy of corrections by considering the operating environment. The operating environment includes, but is not limited to, hardware configuration and network environment. Some or all of the above-described processes in the correction unit may be performed using AI, for example, or without AI. For example, the correction unit can input AI-Agent operating environment data into a generating AI and have the generating AI perform the correction accuracy improvement.

[0089] The editing unit can estimate the user's emotions and adjust how the edits are displayed based on the estimated emotions. For example, if the user is feeling anxious, the editing unit can display detailed edits. If the user is feeling at ease, the editing unit can display concise edits. If the user is in a hurry, the editing unit can display concise edits. By adjusting how the edits are displayed based on the user's emotions, a more appropriate display becomes possible. The method of displaying the edits includes, but is not limited to, the display format and timing of display. Emotion estimation is achieved using an emotion estimation function, for example, an emotion engine or a generative AI. The generative AI is a text generation AI (e.g., LLM) or a multimodal generation AI, but is not limited to these examples. Some or all of the above processing in the editing unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the editing unit can input user emotion data into a generative AI and have the generative AI perform emotion estimation.

[0090] The modification unit can collect user feedback regarding the AI-Agent's operation during the modification process and incorporate it into the modification. For example, the modification unit can collect user feedback and incorporate it into the modification. The modification unit can analyze user feedback and adjust the modification method. The modification unit can improve the accuracy of the modification based on user feedback. Thus, by collecting user feedback, it can be incorporated into the modification. User feedback includes, but is not limited to, the format of the feedback and the tools used to collect it. Some or all of the above-described processes in the modification unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the modification unit can input user feedback into a generative AI and have the generative AI incorporate it into the modification.

[0091] The correction unit can improve the accuracy of corrections by referring to relevant literature on the operation of the AI-Agent during the correction process. The correction unit can improve the accuracy of corrections by referring to relevant literature, for example. The correction unit can adjust the correction method based on the relevant literature. The correction unit can acquire relevant literature in real time and dynamically adjust the accuracy of corrections. As a result, the accuracy of corrections is improved by referring to relevant literature. Relevant literature includes, but is not limited to, literature databases and referencing methods. Some or all of the above processing in the correction unit may be performed using, for example, a generative AI, or without a generative AI. For example, the correction unit can input relevant literature into a generative AI and have the generative AI perform the correction accuracy improvement.

[0092] The service provider can estimate the user's emotions and determine the priority of the information to be provided based on the estimated emotions. For example, if the user is feeling anxious, the service provider can prioritize providing important information. If the user is feeling at ease, the service provider can prioritize providing normal information. If the user is in a hurry, the service provider can prioritize providing urgent information. This allows for more appropriate information provision by prioritizing information based on the user's emotions. Information prioritization includes, but is not limited to, criteria for evaluating importance and methods for prioritizing. Emotion estimation is achieved using an emotion estimation function, for example, an emotion engine or generative AI. Generative AI includes, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the service provider may be performed using, for example, a generative AI, or not using a generative AI. For example, the service provider can input user emotion data into a generative AI and have the generative AI perform emotion estimation.

[0093] The delivery unit can select the optimal delivery method by referring to the AI-Agent's past delivery history at the time of delivery. The delivery unit can, for example, select the optimal delivery method based on past delivery history. The delivery unit can analyze past delivery history and adjust the delivery method. The delivery unit can improve the accuracy of delivery by referring to past delivery history. This allows the optimal delivery method to be selected by referring to past delivery history. The delivery history includes, but is not limited to, the format of the history and the method of referencing. Some or all of the above processing in the delivery unit may be performed using AI, for example, or without using AI. For example, the delivery unit can input past delivery history into a generating AI and have the generating AI select the optimal delivery method.

[0094] The delivery unit can improve the accuracy of deliveries by considering the operating environment of the AI-Agent at the time of delivery. For example, if the network conditions are unstable, the delivery unit can increase the frequency of deliveries to improve accuracy. If the server load is high, the delivery unit can improve accuracy by focusing on important delivery items. The delivery unit can dynamically adjust the accuracy of deliveries in response to changes in the operating environment. This improves the accuracy of deliveries by considering the operating environment. The operating environment includes, but is not limited to, hardware configuration and network environment. Some or all of the above processing in the delivery unit may be performed using AI, for example, or without AI. For example, the delivery unit can input AI-Agent operating environment data into a generating AI and have the generating AI perform the task of improving the accuracy of deliveries.

[0095] The service provider can estimate the user's emotions and adjust how the content is displayed based on the estimated emotions. For example, if the user is feeling anxious, the service provider can display detailed content. If the user is feeling at ease, the service provider can display concise content. If the user is in a hurry, the service provider can display content that gets straight to the point. By adjusting how the content is displayed based on the user's emotions, a more appropriate display becomes possible. The method of displaying the content includes, but is not limited to, the display format and timing of display. Emotion estimation is achieved using an emotion estimation function, for example, an emotion engine or a generative AI. The generative AI is a text generation AI (e.g., LLM) or a multimodal generation AI, but is not limited to these examples. Some or all of the above processing in the service provider may be performed using, for example, a generative AI, or not using a generative AI. For example, the service provider can input user emotion data into a generative AI and have the generative AI perform emotion estimation.

[0096] The service provider can collect user feedback regarding the operation of the AI-Agent at the time of delivery and reflect it in the delivered content. The service provider can, for example, collect user feedback and reflect it in the delivered content. The service provider can analyze user feedback and adjust the delivery method. The service provider can improve the accuracy of the delivery based on user feedback. This allows the service provider to reflect user feedback by collecting it. User feedback includes, but is not limited to, the format of the feedback and the tools used to collect it. Some or all of the above processing in the service provider may be performed using, for example, a generative AI, or not using a generative AI. For example, the service provider can input user feedback into a generative AI and have the generative AI perform the action of reflecting it in the delivered content.

[0097] The delivery unit can improve the accuracy of its delivery by referring to relevant literature on the operation of the AI-Agent at the time of delivery. The delivery unit can improve the accuracy of its delivery by referring to relevant literature. The delivery unit can adjust the delivery method based on the relevant literature. The delivery unit can acquire relevant literature in real time and dynamically adjust the accuracy of its delivery. As a result, the accuracy of its delivery is improved by referring to relevant literature. Relevant literature includes, but is not limited to, literature databases and referencing methods. Some or all of the above processing in the delivery unit may be performed using, for example, a generative AI, or not using a generative AI. For example, the delivery unit can input relevant literature into a generative AI and have the generative AI perform the improvement of the accuracy of its delivery.

[0098] The system according to the embodiment is not limited to the example described above, and various modifications are possible, for example, as follows.

[0099] The monitoring unit not only monitors the operation of the AI-Agent, but can also collect user feedback on the AI-Agent's operation in real time and reflect it in the monitoring results. For example, if a user is dissatisfied with the AI-Agent's operation, they can immediately send that feedback to the monitoring unit, which can then adjust its monitoring focus based on that feedback. Conversely, if a user is satisfied with the AI-Agent's operation, the monitoring frequency can be reduced based on that feedback, saving system resources. Furthermore, by analyzing user feedback, areas for improvement in the AI-Agent's operation can be identified, thereby improving the accuracy of monitoring. In this way, by collecting user feedback in real time and reflecting it in the monitoring results, more appropriate monitoring becomes possible.

[0100] The evaluation unit not only evaluates the output results of the generated AI, but can also improve the accuracy of the evaluation by referring to external data related to the operation of the AI-Agent. For example, it can improve the accuracy of the evaluation by referring to the performance of competitors' AI-Agents. It can also adjust the evaluation criteria based on the external data. Furthermore, it can acquire external data in real time and dynamically adjust the accuracy of the evaluation. As a result, the accuracy of the evaluation is improved by referring to external data. External data includes, but is not limited to, data sources and data formats.

[0101] The modification unit not only corrects the AI-Agent's operation based on the evaluation results obtained by the evaluation unit, but can also improve the accuracy of the modifications by considering the AI-Agent's operating environment. For example, if the network condition is unstable, the frequency of modifications can be increased to improve accuracy. Also, if the server load is high, the accuracy of modifications can be improved by focusing on important items. Furthermore, the accuracy of modifications can be dynamically adjusted in response to changes in the operating environment. In this way, the accuracy of modifications is improved by considering the operating environment.

[0102] The service provider can not only provide the corrected AI-Agent functionality to companies, but also collect user feedback on the AI-Agent's operation and incorporate it into the service. For example, it can collect user feedback and incorporate it into the service. It can also analyze user feedback and adjust the service delivery method. Furthermore, it can improve the accuracy of the service based on user feedback. In this way, by collecting user feedback, it can be incorporated into the service delivery.

[0103] The monitoring unit not only monitors the operation of the AI-Agent, but can also improve the accuracy of monitoring by referring to relevant literature on the AI-Agent's operation. For example, it can improve monitoring accuracy by referring to relevant literature. It can also adjust the monitoring method based on the relevant literature. Furthermore, it can acquire relevant literature in real time and dynamically adjust the accuracy of monitoring. As a result, monitoring accuracy is improved by referring to relevant literature.

[0104] The monitoring unit can estimate the user's emotions and adjust the monitoring frequency based on those emotions. For example, if the user is stressed, the monitoring frequency can be increased to detect problems early. Conversely, if the user is relaxed, the monitoring frequency can be decreased to conserve system resources. Furthermore, if the user is in a hurry, the monitoring frequency can be adjusted to focus on important monitoring items. By adjusting the monitoring frequency based on the user's emotions, more appropriate monitoring becomes possible.

[0105] The evaluation unit can estimate the user's emotions and adjust the evaluation criteria based on those emotions. For example, if the user is feeling anxious, strict evaluation criteria can be applied. Conversely, if the user is feeling at ease, normal evaluation criteria can be applied. Furthermore, if the user is in a hurry, rapid evaluation criteria can be applied. By adjusting the evaluation criteria based on the user's emotions, a more appropriate evaluation becomes possible.

[0106] The correction unit can estimate the user's emotions and determine the priority of corrections based on those emotions. For example, if the user is feeling anxious, important corrections can be prioritized. If the user is feeling at ease, regular corrections can be prioritized. Furthermore, if the user is in a hurry, urgent corrections can be prioritized. By prioritizing corrections based on the user's emotions, more appropriate corrections can be made.

[0107] The information delivery system can estimate the user's emotions and prioritize the information to be delivered based on those emotions. For example, if the user is feeling anxious, important information can be prioritized. If the user is feeling at ease, normal information can be prioritized. Furthermore, if the user is in a hurry, highly urgent information can be prioritized. By prioritizing information based on the user's emotions, more appropriate information can be delivered.

[0108] The service provider can estimate the user's emotions and adjust how the service is displayed based on those emotions. For example, if the user is feeling anxious, detailed service information can be displayed. If the user is feeling at ease, concise service information can be displayed. Furthermore, if the user is in a hurry, the service can display only the essential information. By adjusting the service display method based on the user's emotions, a more appropriate display becomes possible.

[0109] The following briefly describes the processing flow for example form 2.

[0110] Step 1: The monitoring unit monitors the operation of the AI-Agent. For example, the monitoring unit monitors the output generated by the AI-Agent in real time, enabling early detection of problems. Specifically, it detects cases such as when the response generated by the AI-Agent contains inappropriate content or when an excessive number of LLM calls occur. Step 2: The evaluation unit evaluates the problems detected by the monitoring unit. For example, the evaluation unit evaluates the output results of the generation AI using the generation AI itself, and considers improvements to the model's performance and prompts based on the results. This allows the quality of the answers generated by the generation AI to be evaluated and areas for improvement to be identified. Step 3: The modification unit modifies the AI-Agent's operation based on the evaluation results obtained by the evaluation unit. For example, the modification unit improves the reliability of the AI-Agent by modifying its operation based on the evaluation results obtained by the evaluation unit. Specifically, if the response generated by the AI-Agent is inappropriate, the modification unit modifies the response and performs the evaluation again using the generating AI. Step 4: The provisioning unit provides the company with the corrected AI-Agent functionality. The provisioning unit facilitates the adoption of the AI-Agent by providing the company with the corrected AI-Agent functionality. Specifically, it is possible to monitor AI-Agents for various purposes, such as AI-Agents that automatically answer inquiries for companies or AI-Agents that suggest travel itineraries.

[0111] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating 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.

[0112] Data generation model 58 is a form of so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> Examples of generative AI include text generation AI, image generation AI, and multimodal generation AI. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images (e.g., still image data or video data). The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference result in one or more data formats from audio data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the specific processing described above using the data generation model 58. The data generation model 58 may be a fine-tuned model that outputs inference results from prompts that do not contain instructions, in which case the data generation model 58 can output inference results from prompts that do not contain instructions. In the data processing device 12, etc., there are multiple types of data generation models 58, and the data generation model 58 includes AI other than generative AI. AI other than generative AI includes, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVMs), k-means clustering, convolutional neural networks (CNNs), recurrent neural networks (RNNs), generative adversarial networks (GANs), or naive Bayes, and can perform various processes, but is not limited to these examples. Also, the AI ​​may be an AI agent. Furthermore, when the processing of each of the above parts is performed by the AI, the processing may be performed by the AI ​​in part or in whole, but is not limited to this example.Furthermore, processing performed by AI, including generative AI, may be replaced with rule-based processing, and rule-based processing may be replaced with processing performed by AI, including generative AI.

[0113] Furthermore, the processing performed by the data processing system 10 described above is carried out by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart device 14, but it may also be carried out by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart device 14. In addition, the specific processing unit 290 of the data processing device 12 acquires or collects information necessary for processing from the smart device 14 or an external device, and the smart device 14 acquires or collects information necessary for processing from the data processing device 12 or an external device.

[0114] Each of the multiple elements described above, including the monitoring unit, evaluation unit, modification unit, and provision unit, is implemented in at least one of the smart device 14 and the data processing unit 12. For example, the monitoring unit is implemented by the control unit 46A of the smart device 14 and monitors the operation of the AI-Agent in real time. The evaluation unit is implemented by the specific processing unit 290 of the data processing unit 12 and evaluates the output results of the generated AI. The modification unit is implemented by the specific processing unit 290 of the data processing unit 12 and modifies the operation of the AI-Agent based on the evaluation results. The provision unit is implemented by the control unit 46A of the smart device 14 and provides the modified operation of the AI-Agent to the company. The correspondence between each unit and the device or control unit is not limited to the example described above and can be changed in various ways.

[0115] [Second Embodiment] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0116] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0117] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN and / or LAN.

[0118] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0119] The microphone 238 receives voice signals from the user and accepts instructions from the user. The microphone 238 captures the voice signals from the user, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0120] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, which captures images of the area around the user (for example, an imaging range defined by a field of view equivalent to the field of vision of a typical healthy person).

[0121] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0122] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing by the processor 28. The storage 32 stores the specific processing program 56.

[0123] The processor 28 reads a 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 acting as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0124] Storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate the user's emotions using the emotion identification model 59 and perform identification processing using the user's emotions. The emotion estimation function (emotion identification function) using the emotion identification model 59 performs various estimations and predictions regarding the user's emotions, including but not limited to these examples. Furthermore, emotion estimation and prediction also include, for example, emotion analysis.

[0125] In the smart glasses 214, specific processing is performed by the processor 46. The storage 50 stores a specific processing program 60. The processor 46 reads the specific processing program 60 from the storage 50 and executes the read specific processing program 60 on the RAM 48. The specific processing is realized by the processor 46 acting as a control unit 46A according to the specific processing program 60 executed on the RAM 48. The smart glasses 214 also have a data generation model 58 and an emotion identification model 59, similar to the data generation model and emotion identification model 59, and can perform processing similar to that of the specific processing unit 290 using these models.

[0126] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains processing results (such as prediction results) using the data generation model 58 by communicating with the server device that has the data generation model 58. Also, the data processing device 12 may be a server device or a terminal device owned by the user (for example, a mobile phone, robot, home appliance, etc.).

[0127] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0128] The data generation model 58 is a so-called generative AI. An example of a data generation model 58 is a generative AI such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and inference data such as audio data representing speech, text data representing text, and image data representing images (e.g., still image data or video data). The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference result in one or more data formats such as audio data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the specific processing described above using the data generation model 58. The data generation model 58 may be a fine-tuned model that outputs inference results from prompts that do not contain instructions, in which case the data generation model 58 can output inference results from prompts that do not contain instructions. In the data processing device 12, etc., there are multiple types of data generation models 58, and the data generation model 58 includes AI other than generative AI. AI other than generative AI includes, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or naive Bayes, and can perform various processes, but is not limited to these examples. Also, the AI ​​may be an AI agent. Furthermore, when the processing of each part described above is performed by the AI, the processing may be performed by the AI ​​in part or in whole, but is not limited to this example. Also, processing performed by an AI including a generative AI may be replaced by rule-based processing, and rule-based processing may be replaced by processing performed by an AI including a generative AI.

[0129] The data processing system 210 according to the second embodiment performs the same processing as the data processing system 10 according to the first embodiment. The processing by the data processing system 210 is performed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart glasses 214, but it may also be performed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart glasses 214. In addition, the specific processing unit 290 of the data processing device 12 acquires or collects information necessary for processing from the smart glasses 214 or an external device, and the smart glasses 214 acquires or collects information necessary for processing from the data processing device 12 or an external device.

[0130] Each of the multiple elements described above, including the monitoring unit, evaluation unit, modification unit, and provision unit, is implemented, for example, in at least one of the smart glasses 214 and the data processing unit 12. For example, the monitoring unit is implemented by the control unit 46A of the smart glasses 214 and monitors the operation of the AI-Agent in real time. The evaluation unit is implemented, for example, by the specific processing unit 290 of the data processing unit 12 and evaluates the output results of the generated AI. The modification unit is implemented, for example, by the specific processing unit 290 of the data processing unit 12 and modifies the operation of the AI-Agent based on the evaluation results. The provision unit is implemented, for example, by the control unit 46A of the smart glasses 214 and provides the modified operation of the AI-Agent to the company. The correspondence between each unit and the device or control unit is not limited to the example described above and can be modified in various ways.

[0131] [Third Embodiment] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0132] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0133] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN and / or LAN.

[0134] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0135] The microphone 238 receives voice signals from the user and accepts instructions from the user. The microphone 238 captures the voice signals from the user, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0136] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, which captures images of the area around the user (for example, an imaging range defined by a field of view equivalent to the field of vision of a typical healthy person).

[0137] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0138] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0139] The processor 28 reads a 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 acting as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0140] Storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate the user's emotions using the emotion identification model 59 and perform identification processing using the user's emotions. The emotion estimation function (emotion identification function) using the emotion identification model 59 performs various estimations and predictions regarding the user's emotions, including but not limited to these examples. Furthermore, emotion estimation and prediction also include, for example, emotion analysis.

[0141] In the headset terminal 314, specific processing is performed by the processor 46. The storage 50 stores a specific program 60. The processor 46 reads the specific program 60 from the storage 50 and executes the read specific program 60 on the RAM 48. The specific processing is realized by the processor 46 acting as a control unit 46A according to the specific program 60 executed on the RAM 48. The headset terminal 314 also has a data generation model 58 and an emotion identification model 59, similar to the data generation model and emotion identification model 59, and can perform processing similar to that of the specific processing unit 290 using these models.

[0142] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains processing results (such as prediction results) using the data generation model 58 by communicating with the server device that has the data generation model 58. Also, the data processing device 12 may be a server device or a terminal device owned by the user (for example, a mobile phone, robot, home appliance, etc.).

[0143] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0144] The data generation model 58 is a so-called generative AI. An example of a data generation model 58 is a generative AI such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and inference data such as audio data representing speech, text data representing text, and image data representing images (e.g., still image data or video data). The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference result in one or more data formats such as audio data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the specific processing described above using the data generation model 58. The data generation model 58 may be a fine-tuned model that outputs inference results from prompts that do not contain instructions, in which case the data generation model 58 can output inference results from prompts that do not contain instructions. In the data processing device 12, etc., there are multiple types of data generation models 58, and the data generation model 58 includes AI other than generative AI. AI other than generative AI includes, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or naive Bayes, and can perform various processes, but is not limited to these examples. Also, the AI ​​may be an AI agent. Furthermore, when the processing of each part described above is performed by the AI, the processing may be performed by the AI ​​in part or in whole, but is not limited to this example. Also, processing performed by an AI including a generative AI may be replaced by rule-based processing, and rule-based processing may be replaced by processing performed by an AI including a generative AI.

[0145] The data processing system 310 according to the third embodiment performs the same processing as the data processing system 10 according to the first embodiment. The processing by the data processing system 310 is performed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the headset terminal 314, but may also be performed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the headset terminal 314. In addition, the specific processing unit 290 of the data processing device 12 acquires or collects information necessary for processing from the headset terminal 314 or an external device, and the headset terminal 314 acquires or collects information necessary for processing from the data processing device 12 or an external device.

[0146] Each of the multiple elements described above, including the monitoring unit, evaluation unit, modification unit, and provision unit, is implemented in at least one of the headset terminal 314 and the data processing unit 12. For example, the monitoring unit is implemented by the control unit 46A of the headset terminal 314 and monitors the operation of the AI-Agent in real time. The evaluation unit is implemented by the specific processing unit 290 of the data processing unit 12 and evaluates the output results of the generated AI. The modification unit is implemented by the specific processing unit 290 of the data processing unit 12 and modifies the operation of the AI-Agent based on the evaluation results. The provision unit is implemented by the control unit 46A of the headset terminal 314 and provides the modified operation of the AI-Agent to the company. The correspondence between each unit and the device or control unit is not limited to the example described above and can be modified in various ways.

[0147] [Fourth Embodiment] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[0148] As shown in Figure 7, the 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.

[0149] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN and / or LAN.

[0150] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[0151] The microphone 238 receives voice signals from the user and accepts instructions from the user. The microphone 238 captures the voice signals from the user, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0152] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS image sensor or CCD image sensor, which captures images of the area around the user (for example, an imaging range defined by a field of view equivalent to the field of vision of a typical healthy person).

[0153] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0154] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive 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 robot 414's emotions can be expressed by controlling these motors. The robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[0155] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0156] The processor 28 reads a 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 acting as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0157] Storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate the user's emotions using the emotion identification model 59 and perform identification processing using the user's emotions. The emotion estimation function (emotion identification function) using the emotion identification model 59 performs various estimations and predictions regarding the user's emotions, including but not limited to these examples. Furthermore, emotion estimation and prediction also include, for example, emotion analysis.

[0158] In robot 414, specific processing is performed by processor 46. A specific program 60 is stored in storage 50. Processor 46 reads the specific program 60 from storage 50 and executes it on RAM 48. The specific processing is achieved by processor 46 acting as a control unit 46A according to the specific program 60 executed on RAM 48. Robot 414 also has data generation model 58 and emotion identification model 59, similar to those of the robot, and can perform processing similar to that of the specific processing unit 290 using these models.

[0159] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains processing results (such as prediction results) using the data generation model 58 by communicating with the server device that has the data generation model 58. Also, the data processing device 12 may be a server device or a terminal device owned by the user (for example, a mobile phone, robot, home appliance, etc.).

[0160] 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 controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0161] The data generation model 58 is a so-called generative AI. An example of a data generation model 58 is a generative AI such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and inference data such as audio data representing speech, text data representing text, and image data representing images (e.g., still image data or video data). The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference result in one or more data formats such as audio data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the specific processing described above using the data generation model 58. The data generation model 58 may be a fine-tuned model that outputs inference results from prompts that do not contain instructions, in which case the data generation model 58 can output inference results from prompts that do not contain instructions. In the data processing device 12, etc., there are multiple types of data generation models 58, and the data generation model 58 includes AI other than generative AI. AI other than generative AI includes, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or naive Bayes, and can perform various processes, but is not limited to these examples. Also, the AI ​​may be an AI agent. Furthermore, when the processing of each part described above is performed by the AI, the processing may be performed by the AI ​​in part or in whole, but is not limited to this example. Also, processing performed by an AI including a generative AI may be replaced by rule-based processing, and rule-based processing may be replaced by processing performed by an AI including a generative AI.

[0162] The data processing system 410 according to the fourth embodiment performs the same processing as the data processing system 10 according to the first embodiment. The processing by the data processing system 410 is performed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the robot 414, but it may also be performed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the robot 414. In addition, the specific processing unit 290 of the data processing device 12 acquires or collects information necessary for processing from the robot 414 or an external device, and the robot 414 acquires or collects information necessary for processing from the data processing device 12 or an external device.

[0163] Each of the multiple elements described above, including the monitoring unit, evaluation unit, modification unit, and provision unit, is implemented in at least one of the robot 414 and the data processing unit 12. For example, the monitoring unit is implemented by the control unit 46A of the robot 414 and monitors the operation of the AI-Agent in real time. The evaluation unit is implemented by the specific processing unit 290 of the data processing unit 12 and evaluates the output results of the generated AI. The modification unit is implemented by the specific processing unit 290 of the data processing unit 12 and modifies the operation of the AI-Agent based on the evaluation results. The provision unit is implemented by the control unit 46A of the robot 414 and provides the modified operation of the AI-Agent to the company. The correspondence between each unit and the device or control unit is not limited to the example described above and can be modified in various ways.

[0164] Furthermore, the emotion identification model 59, acting 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 a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[0165] Figure 9 shows the emotion map 400, in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[0166] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[0167] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[0168] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, and motorcycles, emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated based, for example, on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[0169] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[0170] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[0171] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing method for the specific process may be used, which includes computer 22 and multiple other computers.

[0172] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[0173] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[0174] Furthermore, it is not necessary to store the entirety of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[0175] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[0176] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[0177] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[0178] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[0179] Furthermore, although the above-described examples were divided into four embodiments, some or all of these embodiments may be combined. Also, the smart device 14, smart glasses 214, headset terminal 314, and robot 414 are just examples, and they may be combined, or other devices may be used. Also, although the above-described examples were divided into two embodiments, Embodiment 1 and Embodiment 2, these may be combined.

[0180] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and other things that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[0181] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

[0182] (Note 1) A monitoring unit that monitors the operation of the AI-Agent, An evaluation unit that evaluates the problems detected by the monitoring unit, A modification unit that modifies the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit, The system includes a provisioning unit that provides the operation of the AI-Agent modified by the modification unit to a company. A system characterized by the following features. (Note 2) The aforementioned monitoring unit, Monitor the output generated by the AI ​​agent in real time. The system described in Appendix 1, characterized by the features described herein. (Note 3) The evaluation unit, We evaluate the output of the generative AI using another generative AI, and based on the results, we consider improvements to the model's performance and prompts. The system described in Appendix 1, characterized by the features described herein. (Note 4) The aforementioned modification section is, The AI-Agent's operation is modified based on the evaluation results obtained by the evaluation unit. The system described in Appendix 1, characterized by the features described herein. (Note 5) The aforementioned supply unit is, Providing the corrected AI-Agent behavior to businesses The system described in Appendix 1, characterized by the features described herein. (Note 6) The aforementioned monitoring unit, Detects problems specific to generative AI, such as hallucination and LLM call counts. The system described in Appendix 1, characterized by the features described herein. (Note 7) The aforementioned monitoring unit, It estimates the user's emotions and adjusts the monitoring frequency based on the estimated emotions. The system described in Appendix 1, characterized by the features described herein. (Note 8) The aforementioned monitoring unit, Analyze the AI ​​agent's operational history and change monitoring focus based on specific patterns. The system described in Appendix 1, characterized by the features described herein. (Note 9) The aforementioned monitoring unit, During monitoring, the accuracy of monitoring is improved by considering the operating environment of the AI-Agent. The system described in Appendix 1, characterized by the features described herein. (Note 10) The aforementioned monitoring unit, It estimates the user's emotions and determines the priority of items to monitor based on the estimated user emotions. The system described in Appendix 1, characterized by the features described herein. (Note 11) The aforementioned monitoring unit, During monitoring, the AI-Agent improves monitoring accuracy by referencing external data related to its operation. The system described in Appendix 1, characterized by the features described herein. (Note 12) The aforementioned monitoring unit, During monitoring, user feedback regarding the AI-Agent's operation is collected in real time and reflected in the monitoring results. The system described in Appendix 1, characterized by the features described herein. (Note 13) The evaluation unit, It estimates the user's emotions and adjusts the evaluation criteria based on the estimated user emotions. The system described in Appendix 1, characterized by the features described herein. (Note 14) The evaluation unit, During evaluation, the evaluation algorithm is optimized by referring to past evaluation results of the AI-Agent. The system described in Appendix 1, characterized by the features described herein. (Note 15) The evaluation unit, During evaluation, external data regarding the AI-Agent's operation is referenced to improve the accuracy of the evaluation. The system described in Appendix 1, characterized by the features described herein. (Note 16) The evaluation unit, The system estimates the user's emotions and adjusts how the evaluation results are displayed based on those estimated emotions. The system described in Appendix 1, characterized by the features described herein. (Note 17) The evaluation unit, During the evaluation process, user feedback regarding the AI-Agent's operation will be collected and incorporated into the evaluation results. The system described in Appendix 1, characterized by the features described herein. (Note 18) The evaluation unit, During evaluation, we improve the accuracy of the evaluation by referring to relevant literature on the operation of the AI-Agent. The system described in Appendix 1, characterized by the features described herein. (Note 19) The aforementioned modification section is, It estimates user sentiment and determines the priority of modifications based on the estimated user sentiment. The system described in Appendix 1, characterized by the features described herein. (Note 20) The aforementioned modification section is, During corrections, the AI-Agent's past correction history is referenced to select the optimal correction method. The system described in Appendix 1, characterized by the features described herein. (Note 21) The aforementioned modification section is, When making corrections, we improve the accuracy of the corrections by taking into account the operating environment of the AI-Agent. The system described in Appendix 1, characterized by the features described herein. (Note 22) The aforementioned modification section is, It estimates the user's emotions and adjusts how the corrections are displayed based on those estimated emotions. The system described in Appendix 1, characterized by the features described herein. (Note 23) The aforementioned modification section is, During the revision process, we will collect user feedback regarding the AI-Agent's behavior and incorporate it into the revisions. The system described in Appendix 1, characterized by the features described herein. (Note 24) The aforementioned modification section is, When making corrections, we refer to relevant literature on AI-Agent operation to improve the accuracy of the corrections. The system described in Appendix 1, characterized by the features described herein. (Note 25) The aforementioned supply unit is, It estimates the user's emotions and prioritizes the information provided based on those estimated emotions. The system described in Appendix 1, characterized by the features described herein. (Note 26) The aforementioned supply unit is, When providing the service, the AI-Agent's past service history is referenced to select the most suitable delivery method. The system described in Appendix 1, characterized by the features described herein. (Note 27) The aforementioned supply unit is, When providing the service, we will improve the accuracy of the service by taking into account the operating environment of the AI-Agent. The system described in Appendix 1, characterized by the features described herein. (Note 28) The aforementioned supply unit is, The system estimates the user's emotions and adjusts how the content is displayed based on those estimated emotions. The system described in Appendix 1, characterized by the features described herein. (Note 29) The aforementioned supply unit is, When providing the service, we will collect user feedback on the AI-Agent's operation and incorporate it into the service. The system described in Appendix 1, characterized by the features described herein. (Note 30) The aforementioned supply unit is, When providing data, we will refer to relevant literature on the operation of the AI-Agent to improve the accuracy of the data provided. The system described in Appendix 1, characterized by the features described herein. [Explanation of symbols]

[0183] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots

Claims

1. A monitoring unit that monitors the operation of the AI-Agent, An evaluation unit that evaluates the problems detected by the monitoring unit, A modification unit modifies the operation of the AI-Agent based on the evaluation results obtained by the evaluation unit, The system includes a provisioning unit that provides the operation of the AI-Agent, modified by the modification unit, to a company. A system characterized by the following features.

2. The aforementioned monitoring unit, Monitor the output generated by the AI-Agent in real time. The system according to feature 1.

3. The evaluation unit described above, We evaluate the output of the generative AI using the generative AI itself, and based on the results, we consider improvements to the model's performance and prompts. The system according to feature 1.

4. The aforementioned modification section is, The operation of the AI-Agent is modified based on the evaluation results obtained by the evaluation unit. The system according to feature 1.

5. The aforementioned supply unit is, Providing the corrected AI-Agent behavior to businesses. The system according to feature 1.

6. The aforementioned monitoring unit, Detects problems specific to generative AI, such as hallucination and LLM call counts. The system according to feature 1.

7. The aforementioned monitoring unit, It estimates the user's emotions and adjusts the monitoring frequency based on the estimated emotions. The system according to feature 1.

8. The aforementioned monitoring unit, Analyze the AI ​​agent's operational history and change monitoring focus based on specific patterns. The system according to feature 1.

Images