Information processing system, information processing device, information processing method, and information processing program

Abstract

Predicting and inferring the prompts that the user will input. [Solution] The information processing device 100 includes an execution means that runs a first AI agent which performs processing to predict a prompt input from a user, and a second AI agent which performs inference processing based on the input prompt. The first AI agent predicts the content of the prompt input from the user and outputs the prediction result to the second AI agent, and the second AI agent performs inference processing based on the prompt input from the first AI agent.

Description

Technical Field

[0005]

[0001] The present invention relates to an information processing system, an information processing apparatus, an information processing method, and an information processing program.

Background Art

[0002] The following system is known. This system operates an organization while causing a plurality of generative AIs to converse with each other (for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] As in the conventional system, there are known methods for obtaining an output from an AI by inputting a prompt sentence to the AI, and techniques for utilizing a plurality of AIs for management. In such a case, if it is possible to predict in advance a prompt input by a user using an AI agent and start inference for the predicted prompt in advance, the output for the input can be speeded up. However, conventionally, no techniques for this purpose have been studied.

Means for Solving the Problems

[0005] The information processing system according to the present invention includes a first AI agent that performs processing for predicting a prompt input from a user, and a second AI agent that performs inference processing based on the input prompt. The first AI agent predicts the content of the prompt input from the user and outputs the prediction result to the second AI agent. The second AI agent performs inference processing based on the prompt input from the first AI agent. The first AI agent, upon receiving a prompt from the user, outputs the input prompt to the second AI agent. The second AI agent, if the user prompt input from the first AI agent matches or is equivalent to a prompt previously predicted by the first AI agent, outputs an inference result based on the previously predicted prompt.It is characterized by the following: The information processing device according to the present invention comprises an execution means that executes a first AI agent that performs processing for predicting a prompt input by a user, and a second AI agent that performs inference processing based on the input prompt, wherein the first AI agent predicts the content of the prompt input by the user and outputs the prediction result to the second AI agent, and the second AI agent performs inference processing based on the prompt input from the first AI agent. The first AI agent, upon receiving a prompt from the user, outputs the input prompt to the second AI agent. The second AI agent, if the user prompt input from the first AI agent matches or is equivalent to a prompt previously predicted by the first AI agent, outputs an inference result based on the previously predicted prompt. It is characterized by the following: The information processing method according to the present invention is a method performed on a computer, wherein the execution means has the steps of executing a first AI agent that performs processing for predicting a prompt input from a user, and a second AI agent that performs inference processing based on the input prompt, wherein the first AI agent predicts the content of the prompt input from the user and outputs the prediction result to the second AI agent, and the second AI agent performs inference processing based on the prompt input from the first AI agent. The first AI agent, upon receiving a prompt from the user, outputs the input prompt to the second AI agent. The second AI agent, if the user prompt input from the first AI agent matches or is equivalent to a prompt previously predicted by the first AI agent, outputs an inference result based on the previously predicted prompt. It is characterized by the following: The information processing program according to the present invention is a program that causes a computer to execute the above-described information processing method. [Effects of the Invention]

[0006] According to the present invention, a second AI agent can initiate inference processing based on a prompt predicted by the first AI agent before the user enters a prompt. [Brief explanation of the drawing]

[0007] [Figure 1] This is a block diagram showing the configuration of one embodiment of the information processing device 100. [Figure 2] This diagram illustrates the data flow and processing of prompts and inference results. [Figure 3] This is a flowchart showing the processing flow executed by the information processing device 100. [Modes for carrying out the invention]

[0008] Figure 1 is a block diagram showing the configuration of one embodiment of the information processing device 100 that constitutes the information processing system in this embodiment. The information processing device 100 may be a personal computer or a server, and includes a communication module 101, a control device 102, and a recording device 103.

[0009] The communication module 101 includes a module for connecting the information processing device 100 to a communication line such as a LAN or the Internet, either wirelessly or via a wired connection. In this embodiment, the information processing device 100 can communicate with other devices by connecting to a communication line via the communication module 101.

[0010] The control device 102 consists of a CPU, memory, and other peripheral circuits, and controls the entire information processing device 100. The memory that makes up the control device 102 is a volatile memory such as SDRAM. This memory is used as work memory for the CPU to load programs when executing programs, and as buffer memory for temporarily recording data. For example, data received via the communication module 101 is temporarily recorded in the buffer memory.

[0011] The recording device 103 is a recording device for recording various data stored by the information processing device 100, as well as program data for execution by the control device 102. For example, an HDD (Hard Disk Drive) or SSD (Solid State Drive) may be used. The program data to be recorded in the recording device 103 is provided on a recording medium such as a CD-ROM or DVD-ROM, or via a network. By installing the program data obtained by the user into the recording device 103, the control device 102 becomes able to execute the program.

[0012] In this embodiment, the information processing device 100 performs processing using an AI agent, which will be described later. Here, an AI agent is a system that acts as a substitute (agent) for a human, possesses the ability to learn and make decisions on its own, and integrates various AI technologies for solving complex problems. For example, when a user gives the AI ​​agent a question (prompt), it searches for an answer to the question or generates an answer and outputs an answer to the question. Also, for example, when a user gives the AI ​​agent a task (prompt), it searches for or generates a solution to the task and outputs a solution to the task.

[0013] In this embodiment, an LLM (Large-Scale Language Model) is assumed as an example of an AI agent. However, the AI ​​agent according to the present invention is not limited to an LLM (Large-Scale Language Model).

[0014] In this embodiment, the information processing device 100 uses a first AI agent that has learned data to predict prompts input by the user, and a second AI agent that has learned data related to business information, to propose answers to questions and solutions to problems (prompts) presented by the user. The data related to business information includes, for example, data related to purchase history at stores, etc.

[0015] The first AI agent is a system that uses a large-scale language model, which has learned the content of prompts previously entered by the user, to perform processing to predict what kind of prompt the user will enter. By using this first AI agent, it is possible to predict the prompts that a user accessing the information processing device 100 will enter.

[0016] The second AI agent is a system that uses a large language model trained on data related to business information including data on purchase history to perform processing for continuously thinking about answers to the input prompts. By using this second AI agent, it is possible to continuously think about answers to the input prompts while incorporating new information.

[0017] In this embodiment, the prediction result by the first AI agent is input to the second AI agent. As a result, the second AI agent can continuously think about answers to the prompts based on the prompts predicted by the first AI agent before the user inputs the prompts.

[0018] The first AI agent waits for the input of a prompt from the user and learns the input prompt when the prompt is input by the user. Also, the first AI agent outputs the prompt actually input by the user to the second AI agent.

[0019] The second AI agent starts the inference process triggered by the prompt input from the first AI agent. That is, when the prompt actually input by the user input from the first AI agent matches or is synonymous with the prompt predicted in advance by the first AI agent, the second AI agent outputs the inference result that it has been continuously thinking about in advance to the first AI agent. The first AI agent outputs the inference result output from the second AI agent. As a result, the inference result that the second AI agent has been continuously thinking about in advance can be output at high speed.

[0020] On the one hand, when the prompt actually input by the user input from the first AI agent does not match the prompt predicted in advance by the first AI agent and is not synonymous, the second AI agent makes an inference based on the prompt actually input by the user input from the first AI agent, and outputs the inference result to the first AI agent. The first AI agent outputs the inference result output from the second AI agent. As a result, even if the prompt input by the user is different from the one predicted in advance by the first AI agent, the inference result for the prompt input by the user can be output.

[0021] For example, assume that user A has asked many questions about sales prediction in the past, user B has asked many questions about purchase prediction, and the first AI agent has learned this information. In this case, for user A, the first AI agent predicts that a question about sales prediction will be input, and inputs the content of the predicted question to the second AI agent. The second AI agent can continue to consider the question content input from the first AI agent while incorporating new information. Also, for user B, the first AI agent predicts that a question about purchase prediction will be input, and inputs the content of the predicted question to the second AI agent. The second AI agent can continue to consider the question content input from the first AI agent while incorporating new information.

[0022] In this embodiment, assume that the program for executing the first AI agent and the program for executing the second AI agent are recorded in the recording device 103. Then, by the control device 102 executing the program for executing the first AI agent and the program for executing the second AI agent, the first AI agent and the second AI agent can be activated. As a result, the first AI agent and the second AI agent become available in the information processing device 100.

[0023] The user can utilize the first AI agent and the second AI agent by accessing the information processing device 100 from the user terminal. Figure 2 is a diagram illustrating the data flow and processing content of prompts and inference results in the information processing system of this embodiment. Figure 2 schematically shows the data flow between the information processing device 100 and the user terminal 200, and the data flow between the first AI agent and the second AI agent in the information processing system. The details of the processing performed by the first AI agent and the second AI agent in the information processing device 100 will be explained below using Figure 2.

[0024] When the control device 102 receives access from the user terminal 200, it identifies the user operating the user terminal 200 and inputs information indicating the user's identity to the first AI agent. The method for identifying the user operating the user terminal 200 is not particularly limited, but for example, the user could be identified using login information for the information processing device 100.

[0025] As described above, the first AI agent uses a large-scale language model that has learned from the content of prompts previously entered by the user to predict what kind of prompt the user will enter. The first AI agent then outputs the prediction result, i.e., the predicted prompt, to the second AI agent.

[0026] As mentioned above, the second AI agent uses a large-scale language model trained on business information data, including purchase history data, to perform inference processing while incorporating new information based on predictive prompts input from the first AI agent.

[0027] The control device 102 receives prompt input from the user terminal 200 connected via the communication module 101. For example, the user terminal 200 has a function for inputting prompts to the first AI agent and a function for displaying the inference results output by the first AI agent. For example, the user terminal 200 may accept prompt input to the first AI agent on its screen and display the inference results output by the first AI agent on the screen. Alternatively, the user terminal 200 may accept voice input of prompts to the first AI agent and output the inference results output by the first AI agent as voice. The prompt data entered by the user on the user terminal 200 is transmitted via the communication line. When the control device 102 receives prompt data from the user terminal 200 via the communication module 101, it inputs the received prompt data to the first AI agent.

[0028] The first AI agent learns from the prompts entered by the user. This allows it to learn from the user's input prompts and improve the accuracy of future predictions. The first AI agent also outputs the user's input prompts to the second AI agent.

[0029] If the user prompt input from the first AI agent matches or is equivalent to the first AI agent's prior prediction, the second AI agent outputs an inference result based on the previously predicted prompt, i.e., an inference result that it had been considering in advance, to the first AI agent. The first AI agent outputs the inference result output from the second AI agent. The control device 102 transmits the inference result output from the first AI agent to the user terminal 200 via the communication line. As a result, the inference result is output on the user terminal 200, and the inference result can be presented to the user. In this case, since the second AI agent outputs an inference result based on a previously predicted prompt, it can output the inference result at high speed.

[0030] Meanwhile, if the user prompt input from the first AI agent does not match or is not equivalent to the prior prediction result of the first AI agent, the second AI agent performs inference processing based on the newly input user prompt and outputs the inference result to the first AI agent. The first AI agent outputs the inference result output from the second AI agent. The control device 102 transmits the inference result output from the first AI agent to the user terminal 200 via the communication line. As a result, the inference result is output on the user terminal 200, and the inference result can be presented to the user.

[0031] Figure 3 is a flowchart showing the processing flow executed by the information processing device 100 in this embodiment. The processing shown in Figure 3 is executed by the control device 102 as a program that is started when a prompt is input from the user terminal 200. In Figure 3, it is assumed that the program for executing the first AI agent and the program for executing the second AI agent have already been executed by the control device 102.

[0032] In step S10, the control device 102 inputs the question and task prompts received from the user terminal 200 to the first AI agent, as described above. Then, the process proceeds to step S20.

[0033] In step S20, the control device 102 outputs the output result from the first AI agent to the user terminal 200, as described above. After that, the process ends.

[0034] According to the embodiments described above, the following effects and advantages can be obtained. (1) The control device 102 executes a first AI agent that performs processing to predict prompts to be input by the user, and a second AI agent that performs inference processing based on the input prompt. The first AI agent predicts the content of the prompt to be input by the user and outputs the prediction result to the second AI agent, and the second AI agent performs inference processing based on the prompt input by the first AI agent. As a result, the second AI agent can start inference processing based on the prompt predicted by the first AI agent before the user inputs a prompt.

[0035] (2) When the first AI agent receives a prompt from the user, it outputs the input prompt to the second AI agent. The second AI agent outputs an inference result based on the previously predicted prompt if the user prompt input from the first AI agent matches or is equivalent to a prompt previously predicted by the first AI agent. This allows the second AI agent to output the results of its prior inference at high speed.

[0036] (3) The second AI agent is configured to perform inference processing based on the newly input user prompt if the user prompt input from the first AI agent does not match or is not synonymous with the prompt previously predicted by the first AI agent, and outputs the inference result. This allows the AI ​​agent to output an inference result for the user prompt even if the user prompt is different from the one previously predicted by the first AI agent.

[0037] (4) The second AI agent outputs the inference result to the first AI agent, the first AI agent outputs the inference result output from the second AI agent, and the control device 102 presents the inference result output from the first AI agent to the user. This makes it possible to present the inference result to the user.

[0038] (5) The second AI agent is a system that utilizes a large-scale language model trained on data related to business information. This allows the second AI agent to receive questions from users regarding business information and provide inference results.

[0039] (6) The data related to business information now includes data related to purchase history. This allows the second AI agent to receive questions from the user regarding purchase history and provide inference results.

[0040] —Revised Version— Furthermore, the contents described in the above-mentioned embodiment can also be modified as follows.

[0041] (1) In the embodiments described above, the data relating to management information was described in an example that included data relating to purchase history at stores, etc. However, the data relating to management information is not limited to data relating to purchase history, as long as it is data to which the present invention can be applied. For example, the data relating to management information may include data such as purchase history data, sales floor data, sales promotion data, sales floor sales promotion data, inventory data, product master data, business manual / policy guide data, in-store congestion data, and data that combines these.

[0042] (2) In the above-described embodiment, an example was given in which the program for executing the first AI agent and the program for executing the second AI agent are recorded in the recording device 103 of the information processing device 100. However, the program for executing the first AI agent and the program for executing the second AI agent may be recorded in an external recording device.

[0043] (3) In the embodiments described above, the prediction target of the first AI agent is assumed to be the content of a prompt entered by the user, and the first AI agent waits for prompt input from the user and inputs this to the second AI agent, thereby triggering the activation of the second AI agent. However, the activation triggers given to the second AI agent are not limited to prompt input. For example, the second AI agent may perform inference activation when an activation trigger is provided based on time-dependent, location-dependent, operation-triggered, inactivity-triggered, past history-dependent, or external event-dependent triggers.

[0044] (4) In the embodiments described above, the information processing device 100 uses a first AI agent that has learned data for predicting prompts input by the user and a second AI agent that has learned data related to management information to propose answers to questions and solutions to problems (prompts) presented by the user. However, the first AI agent and the second AI agent may be a single AI agent. That is, an AI agent is prepared that performs the prediction processing performed by the first AI agent and the inference processing performed by the second AI agent in the embodiments described above. This AI agent learns data for predicting prompts input by the user and also learns data related to management information to predict prompts input by the user and continues to think of answers to prompts based on the predicted prompts. The AI ​​agent then outputs the inference result it had been thinking about in advance if the prompt actually input by the user matches or is synonymous with a previously predicted prompt. If the prompt actually input by the user does not match or is not synonymous with a previously predicted prompt, the AI ​​agent performs inference based on the prompt actually input by the user and outputs the inference result. This enables the provision of an AI agent having steps for performing processing to predict an activation trigger (prompt) and for performing inference processing based on the activation trigger (prompt), wherein the AI ​​agent can predict the content of the activation trigger and perform inference processing based on the predicted activation trigger.

[0045] Furthermore, the present invention is not limited in any way to the configurations in the embodiments described above, as long as the characteristic functions of the present invention are not impaired. [Explanation of Symbols]

[0046] 100 Information Processing Devices 101 Communication Module 102 Control device 103 Recording device 200 user terminals

Claims

1. An information processing system comprising a first AI agent that performs processing to predict prompts input by a user, and a second AI agent that performs inference processing based on the input prompt, The first AI agent predicts the content of the prompt input by the user and outputs the prediction result to the second AI agent. The second AI agent performs inference processing based on the prompt input from the first AI agent. When the first AI agent receives a prompt from the user, it outputs the input prompt to the second AI agent. The information processing system is characterized in that the second AI agent outputs an inference result based on a previously predicted prompt when the user prompt input from the first AI agent matches or is equivalent to a prompt previously predicted by the first AI agent.

2. In the information processing system described in claim 1, The information processing system is characterized in that, if the user prompt input from the first AI agent does not match or is not synonymous with a prompt previously predicted by the first AI agent, the second AI agent performs inference processing based on the newly input user prompt and outputs the inference result.

3. In the information processing system according to claim 1 or 2, The second AI agent outputs the inference result to the first AI agent. The information processing system is characterized in that the first AI agent outputs the inference results output from the second AI agent.

4. In the information processing system described in claim 3, An information processing system further comprising presentation means for processing the inference results output from the first AI agent to the user.

5. In the information processing system described in claim 1, The aforementioned second AI agent is an information processing system characterized by utilizing a large-scale language model that has learned from data related to business information.

6. In the information processing system described in claim 5, An information processing system characterized in that the data relating to the aforementioned management information includes data relating to purchase history.

7. An information processing device comprising an execution means for executing a first AI agent that performs processing for predicting prompts input by a user, and a second AI agent that performs inference processing based on the input prompt, The first AI agent predicts the content of the prompt input by the user and outputs the prediction result to the second AI agent. The second AI agent performs inference processing based on the prompt input from the first AI agent. When the first AI agent receives a prompt from the user, it outputs the input prompt to the second AI agent. The second AI agent is characterized in that, if the user prompt input from the first AI agent matches or is equivalent to a prompt previously predicted by the first AI agent, it outputs an inference result based on the previously predicted prompt.

8. In the information processing apparatus according to claim 7, The information processing device is characterized in that, if the user prompt input from the first AI agent does not match or is not equivalent to a prompt previously predicted by the first AI agent, the second AI agent performs inference processing based on the newly input user prompt and outputs the inference result.

9. In the information processing apparatus according to claim 7 or 8, The second AI agent outputs the inference result to the first AI agent. The first AI agent is characterized by outputting the inference result output from the second AI agent.

10. In the information processing apparatus according to claim 9, An information processing device further comprising presentation means for processing the inference results output from the first AI agent to the user.

11. In the information processing apparatus according to claim 7, The information processing device is characterized in that the second AI agent is a system that utilizes a large-scale language model that has learned data related to business information.

12. In the information processing apparatus according to claim 11, The information processing device is characterized in that the data relating to the aforementioned management information includes data relating to purchase history.

13. An information processing method performed on a computer, comprising the steps of: executing a first AI agent that performs processing for predicting a prompt input by a user; and executing a second AI agent that performs inference processing based on the input prompt, The first AI agent predicts the content of the prompt input by the user and outputs the prediction result to the second AI agent. The second AI agent performs inference processing based on the prompt input from the first AI agent. When the first AI agent receives a prompt from the user, it outputs the input prompt to the second AI agent. The information processing method is characterized in that the second AI agent outputs an inference result based on a previously predicted prompt when the user prompt input from the first AI agent matches or is equivalent to a prompt previously predicted by the first AI agent.

14. In the information processing method described in claim 13, The information processing method is characterized in that, if the user prompt input from the first AI agent does not match or is not synonymous with a prompt previously predicted by the first AI agent, the second AI agent performs inference based on the newly input user prompt and outputs the inference result.

15. In the information processing method described in claim 13, The second AI agent outputs the inference result to the first AI agent. An information processing method characterized in that the first AI agent outputs the inference results output from the second AI agent.

16. In the information processing method described in claim 15, An information processing method characterized in that the presentation means further comprises the step of processing for presenting the inference results output from the first AI agent to the user.

17. In the information processing method described in claim 13, The information processing method is characterized in that the second AI agent is a system that utilizes a large-scale language model that has learned data related to business information.

18. In the information processing method described in claim 17, An information processing method characterized in that the data relating to the aforementioned management information includes data relating to purchase history.

19. An information processing program for causing a computer to perform the method described in any one of claims 13 to 18.

Images