Business support system, business support method, and program

Abstract

To improve the convenience for a user.SOLUTION: An input data acquisition unit (2001) of a business support system (1) acquires input data indicating user input. From among a plurality of business support programs specialized for a specific task in business support, a selection unit (2002) can select a business support program that is specialized for a task corresponding to the input data, on the basis of an artificial intelligence (AI) that is capable of selecting the business support program. When the business support program is selected by the selection unit, a task processing unit (2003) processes the task that is in accordance with the input data on the basis of the business support program.SELECTED DRAWING: Figure 4

Description

Technical Field

[0001] The present disclosure relates to a business support system, a business support method, and a program.

Background Art

[0002] Conventionally, technologies for supporting a user's business based on AI (Artificial Intelligence) have been studied. For example, Patent Document 1 describes a large language model such as BERT (Bidirectional Encoder Representations from Transformers), which is an example of such AI. For example, a large language model such as BERT has high versatility and can process various tasks based on input data indicating a user's input.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, since a highly versatile AI such as that in Patent Document 1 is not specialized for a specific task, the accuracy of task processing may not be sufficient. On the other hand, if a business support system tries to prepare an AI specialized for a certain specific task, it is necessary to prepare a large number of AIs in order to handle various tasks. The user has to select an AI specialized for a specific task from among the large number of AIs and have the task processed. Therefore, the conventional technology cannot sufficiently improve the convenience for the user. This is true not only for AI but also for business support programs in general that support a user's business.

[0005] One of the objects of the present disclosure is to improve the convenience for the user.

Means for Solving the Problem

[0006] A business support system according to one aspect of the present disclosure includes an input data acquisition unit that acquires input data indicating a user's input, and among a plurality of business support programs specialized for specific tasks in business support, an AI (Artificial Intelligence) capable of selecting the business support program specialized for the task corresponding to the input data based on the input data, a selection unit capable of selecting the business support program, and a task processing unit that processes the task corresponding to the input data based on the business support program when the business support program is selected by the selection unit.

Effect of the Invention

[0007] According to the present disclosure, user convenience can be enhanced.

Brief Description of the Drawings

[0008]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Embodiment for Carrying Out the Invention

[0009] [1. Hardware Configuration] An example of an embodiment of a business support system, a business support method, and a program according to the present disclosure will be described. FIG. 1 is a diagram showing an example of the hardware configuration of the business support system. For example, the business support system 1 includes a learning terminal 10, a server 20, and a user terminal 30. Each of the learning terminal 10, the server 20, and the user terminal 30 is connected to a network N such as the Internet or a LAN.

[0010] The learning terminal 10 is a computer that performs learning of AI (Artificial Intelligence) described later. For example, the learning terminal 10 is a personal computer, a tablet terminal, or a smartphone. For example, the learning terminal 10 includes a control unit 11, a storage unit 12, a communication unit 13, an operation unit 14, and a display unit 15. The control unit 11 includes at least one processor. The storage unit 12 includes at least one of a volatile memory such as a RAM and a non-volatile memory such as a flash memory. The communication unit 13 includes at least one of a communication interface for wired communication and a communication interface for wireless communication. The operation unit 14 is an input device such as a mouse or a touch panel. The display unit 15 is a liquid crystal or organic EL display.

[0011] The server 20 is a server computer. For example, the server 20 includes a control unit 21, a storage unit 22, and a communication unit 23. The hardware configurations of the control unit 21, the storage unit 22, and the communication unit 23 may be the same as those of the control unit 11, the storage unit 12, and the communication unit 13, respectively.

[0012] The user terminal 30 is a user's computer. For example, the user terminal 30 is a personal computer, a tablet terminal, a smartphone, or a wearable terminal. For example, the user terminal 30 includes a control unit 31, a storage unit 32, a communication unit 33, an operation unit 34, and a display unit 35. The hardware configurations of the control unit 31, the storage unit 32, the communication unit 33, the operation unit 34, and the display unit 35 may be the same as those of the control unit 11, the storage unit 12, the communication unit 13, the operation unit 14, and the display unit 15, respectively.

[0013] Note that the programs stored in the storage units 12, 22, and 32 may be supplied via the network N. The hardware configurations of each of the learning terminal 10, the server 20, and the user terminal 30 are not limited to the example of FIG. 1. For example, at least one of the learning terminal 10, the server 20, and the user terminal 30 may include at least one of a reading unit (e.g., a memory card slot) that reads a computer-readable information storage medium and an input / output unit (e.g., a USB terminal) for directly connecting to an external device. The program stored in the information storage medium may be supplied to at least one of the learning terminal 10, the server 20, and the user terminal 30 via at least one of the reading unit and the input / output unit.

[0014] Also, the business support system 1 may include at least one computer. The computer included in the business support system 1 is not limited to the example of FIG. 1. For example, the business support system 1 may include only the learning terminal 10 and the server 20. In this case, the user terminal 30 exists outside the business support system 1. The business support system 1 may include only the server 20. In this case, the learning terminal 10 and the user terminal 30 exist outside the business support system 1. The business support system 1 may include the server 20 and another server computer.

[0015] [2. Overview of the Business Support System] In the present embodiment, the business support system 1 has a business support function for supporting a user's business. For example, the business support function may be a communication function for a user to communicate with other users, a schedule management function for managing a user's schedule, a file management function for managing a user's files, or a mail management function for managing a user's mails. The business support function may be other known functions.

[0016] For example, the business support system 1 may provide users with cloud-based or on-premises groupware. The business support system 1 may provide users with services that are not classified as groupware but support business operations. When a user logs in to the business support system 1, the user terminal 30 causes the display unit 35 to display a business support screen for the user to use the business support function. In the present embodiment, a case where the business support screen is displayed on the browser of the user terminal 30 is taken as an example.

[0017] FIG. 2 is a diagram showing an example of a business support screen. In the example of FIG. 2, a business support screen SC for a user to use a communication function, which is an example of a business support function, is shown. For example, the user terminal 30 causes the business support screen SC to display a thread selected by the user from among a plurality of threads created in the business support system 1. Posts made to the thread are arranged in chronological order on the business support screen SC. The user can make a new post to the thread or make a reply post to an existing post.

[0018] As in the upper business support screen SC of FIG. 2, the user can enter a new post in the input form F10. For example, the user can mention another user by specifying the other user. A mention is a notification to another user. In the example of FIG. 2, the user can mention another user by entering the information of another user (for example, the name of another user) after a specific symbol (for example, @). The mechanism of the mention may be the same as a known mechanism. The user can also mention the AI in the same way as the mention to another user.

[0019] AI is a program with artificial intelligence that supports a user's business. Although there are various definitions of AI, the AI in this embodiment may be an AI defined by various known definitions. The AI may be an AI called generative AI or conversational AI. For example, the AI may be a large language model, a machine learning model not classified as a large language model, a program called a bot, or other programs. Although there are various definitions of machine learning, the machine learning in this embodiment may be a machine learning defined by various known definitions. The machine learning may be any of supervised learning, semi-supervised learning, or unsupervised learning.

[0020] In this embodiment, a case where a large language model corresponds to the AI is taken as an example. For example, a user can mention the AI by inputting a character string (e.g., bot) indicating the AI instead of the information of another user after a specific symbol (e.g., @). When the user inputs a post, the user terminal 30 transmits input data indicating the post input by the user to the server 20.

[0021] For example, when the server 20 receives the input data from the user terminal 30, it determines whether a mention of the AI has been made based on the input data. If the server 20 determines that no mention of the AI has been made, it reflects the user's post in the thread. If the server 20 determines that a mention of the AI has been made, it causes the AI to process a task corresponding to the input data.

[0022] A task is the content of a process executed by a program such as AI. A task can also be the content of the output from a program such as AI. The task can be any task related to business support. For example, the task can be machine translation, summary creation, research (providing knowledge), schedule adjustment, generation of text such as emails or reports, image creation, application generation, database configuration, classification of input data, or other tasks. The business support system 1 can process a plurality of such tasks as examples. The processing of tasks is also an example of the business support functions of the business support system 1.

[0023] A task corresponding to input data is a task determined based on the input indicated by the input data (in the example of FIG. 2, a post to a thread). A task corresponding to input data can also be a task requested by the user who made the input indicated by the input data. For example, when the AI is a generative AI, the task corresponding to the input data is a task for the AI to generate content such as an image or text based on the input data. When the AI is a conversational AI, the task corresponding to the input data is a task for generating response data indicating an answer from the AI based on the input indicated by the input data.

[0024] In the present embodiment, a plurality of AIs specialized for specific tasks in business support are prepared. An AI is an example of a business support program for supporting a user's business. As in the modification example described later, the business support program may be a program not classified as an AI, but in the present embodiment, the case where the AI corresponds to the business support program is taken as an example. The server 20 causes an AI specialized for the task corresponding to the input data among the plurality of pre-prepared AIs to process the task.

[0025] In the example of FIG. 2, the input data shows a post such as "I will explain the popular programming language to the department head at the regular meeting next week. Please teach me the popular programming language." Since the user requests the AI to research popular programming, the task corresponding to the input data is a research task. The server 20 causes an AI specialized in the research task among a plurality of pre-prepared AIs to process the research task. The server 20 transmits data indicating the processing result of the task output by the AI to the user terminal 30. The user terminal 30 displays the answer from the AI based on the data, as shown in the business support screen SC at the lower side of FIG. 2.

[0026] As described above, the server 20 causes an AI specialized in the task corresponding to the input data among the plurality of AIs to process the task. For example, assume that the user selects by himself / herself an AI specialized in the task corresponding to the input data from among the plurality of AIs, and the server 20 causes the AI selected by the user to process the task. In this case, since the user needs to select an AI, it is troublesome for the user. Furthermore, since the user does not always select an appropriate AI, there is also a possibility that the task is not processed appropriately.

[0027] Therefore, in the present embodiment, an AI capable of selecting the AI is prepared in advance separately from the AI specialized in a specific task. Hereinafter, the AI that performs processing specialized in a specific task is referred to as a child AI. The child AI can also be referred to as a specialized type AI. The AI capable of selecting the child AI is referred to as a parent AI. The parent AI can also be referred to as a selection type AI. In the present embodiment, the case where the parent AI only performs the task of selecting the child AI is taken as an example, but as in a modification example described later, the parent AI may be a general-purpose AI capable of processing various tasks. That is, the parent AI does not have to be specialized only in the task of selecting the child AI and may be capable of handling various tasks.

[0028] FIG. 3 is a diagram showing an example of the relationship between the parent AI and the child AIs. In the example of FIG. 3, child AI 201A specialized for the task of machine translation, child AI 201B specialized for the task of summary creation, child AI 201C specialized for the task of research, and child AI 201D specialized for the task of schedule adjustment are prepared. Hereinafter, when not particularly distinguishing between child AIs 201A to 201D, they are simply referred to as child AI 201. Further, when not particularly distinguishing between the parent AI 200 and the child AI 201, after omitting the reference signs, they are simply referred to as AI.

[0029] For example, the server 20 inputs the input data received from the user terminal 30 to the parent AI 200. Assume that the parent AI 200 has been pre-trained so that it can select an appropriate child AI 201. The learning method of the parent AI 200 will be described later. Depending on the content of the input data, the parent AI 200 may not need to select a child AI 201. In the example of the input data in FIG. 2, the parent AI 200 estimates from the content of the input data that the child AI 201C specialized for the task of research is appropriate, and selects the child AI 201C.

[0030] For example, the server 20 inputs the input data received from the user terminal 30 to the child AI 201C selected by the parent AI 200. The child AI 201C processes the task of research based on the input data, and outputs output data indicating the result of the research. In the example of FIG. 2, the output data output from the child AI 201C shows an answer such as "The popularity of programming languages is always changing, but at the time of 2023, several languages are attracting attention. For example, the following five languages are...".

[0031] In this embodiment, the child AI 201 selected by the parent AI 200 varies depending on the content of the input data. For example, assume that the user inputs a post including a sentence described in Japanese and a sentence such as "Please translate the above sentence into English." In this case, the parent AI 200 estimates from the content of the input data that the child AI 201A specialized in the task of machine translation is appropriate, and selects the child AI 201A. The server 20 causes the child AI 201A selected by the parent AI 200 to process the task of machine translation based on the input data. The child AI 201A processes the task of machine translation by translating the Japanese sentence indicated by the input data into English.

[0032] For example, assume that the user inputs a post including a sentence of a certain length and a sentence such as "Please create a summary of the above sentence." In this case, the parent AI 200 estimates from the content of the input data that the child AI 201B specialized in the task of summary creation is appropriate, and selects the child AI 201B. The server 20 causes the child AI 201B selected by the parent AI 200 to process the task of summary creation based on the input data. The child AI 201B processes the task of summary creation by creating a summary of the sentence indicated by the input data.

[0033] For example, assume that the user inputs a post including candidate dates for a meeting and a sentence such as "Please check if my schedule is available." In this case, the parent AI 200 estimates from the content of the input data that the child AI 201D specialized in the task of schedule adjustment is appropriate, and selects the child AI 201D. The server 20 causes the child AI 201D selected by the parent AI 200 to process the task of schedule adjustment based on the input data. The child AI 201D processes the task of schedule adjustment by referring to the schedule data of the user registered in the business support system 1 and checking whether the candidate date indicated by the input data is available.

[0034] As described above, in the business support system 1 of the present embodiment, the parent AI 200 can select a child AI 201 specialized for a task corresponding to the input data based on the input data indicating the post input by the user. When the child AI 201 is selected by the parent AI 200, the child AI 201 processes the task corresponding to the input data. As a result, the business efficiency of the user is improved, so the business support system 1 can enhance the convenience of the user. Hereinafter, the details of the business support system 1 will be described.

[0035] [Functions realized by the business support system] FIG. 4 is a diagram showing an example of functions realized by the business support system 1.

[0036] [3-1. Functions realized by the learning terminal] For example, the learning terminal 10 includes a data storage unit 1000 and a learning unit 1001. The data storage unit 1000 is realized by the storage unit 12. The learning unit 1001 is realized by the control unit 11.

[0037] [Data storage unit] The data storage unit 1000 stores data necessary for the learning of at least one of the parent AI 200 and the child AI 201. For example, the data storage unit 1000 stores the parent AI 200. In the present embodiment, a case where the parent AI 200 is a large-scale language model based on a transformer such as GPT (Generative Pre-trained Transformer) or BERT (Bidirectional Encoder Representations from Transformers) is taken as an example. The parent AI 200 may be a large-scale language model of other methods other than transformers. The parent AI 200 may be a machine learning model not classified as a large-scale language model. For example, the parent AI 200 may be an adversarial generation network, a neural network, or a support vector machine.

[0038] Note that since the selection of the appropriate child AI 201 can also be regarded as the classification (labeling) of input data, the parent AI 200 may be a machine learning model capable of classifying input data. Further, the parent AI 200 may be a program that is not classified as a machine learning model. For example, the parent AI 200 may be a rule-based program that indicates the relationship between input data and an appropriate child AI 201. The rule-based program defines rules that are determined based on all or part of the input indicated by the input data. The rule-based program describes program code indicating that the child AI 201 associated with the rule is appropriate when a certain rule is satisfied. When the parent AI 200 is a rule-based program, since learning is not performed, the business support system 1 may not include the learning terminal 10.

[0039] In this embodiment, since the parent AI 200 is a machine learning model exemplified by a transformer-based large language model, the parent AI 200 includes parameters adjusted by learning and a program indicating processes such as the calculation of embedded representations. Further, in this embodiment, an example is given of a case where a business operator (for example, a company that developed groupware) that provides the business support system 1 to a user creates the parent AI 200 by performing retraining (for example, fine-tuning, transfer learning, or distillation) of a pre-trained large language model provided by another business operator.

[0040] Note that the program and parameters of the parent AI 200 may be various known programs and parameters. The parameters of the parent AI 200 are referenced by the program of the parent AI 200. For example, the parameters are at least one of a weight coefficient and a bias. The program of the parent AI 200 includes code indicating the internal processing of the parent AI 200. For example, the program of the parent AI 200 includes an intermediate layer that calculates an embedded representation and generates data for the output layer, and an output layer that performs a final output based on the data.

[0041] For example, if the parent AI 200 is a transformer-based large language model, the program of the parent AI 200 includes processes of splitting the text indicated by the input data into a plurality of tokens, calculating the embedded representation of each token, making predictions according to the order of the embedded representations, and outputting according to this series of processes. When the parent AI 200 is another machine learning model, the program of the parent AI 200 only needs to indicate the processes adopted as the other machine learning model. The data storage unit 1000 stores a training database DB in which training data to be learned by the learning unit 1001 described later to the parent AI 200 is stored.

[0042] FIG. 5 is a diagram showing an example of the training database DB. In the present embodiment, a case where learning of supervised learning is performed is taken as an example. For example, the training data includes an input part input to the AI during learning and an output part that is the correct answer during learning. In the example of FIG. 5, the input part of the training data is the input data for training. The output part of the training data is child AI identification data for identifying the child AI 201 (child AI 201 specialized for a task according to the training input data) that is the correct answer. When the parent AI 200 is created by re-learning a pre-trained large language model, the training data is data for re-learning. The learning in the present embodiment includes re-learning.

[0043] The input data for training indicates the input for training. For example, the input data for training indicates a text (character string) for training. As in the example of FIG. 2, when the input data indicating the post of the thread is analyzed by the parent AI 200, the input data for training is the post that the learning unit 1001 causes the parent AI 200 to learn. The input data for training may indicate other information other than text. For example, the input data for training may indicate a file, a reaction, an image displayed on the business support screen SC, or other information.

[0044] The child AI identification data is data that enables the business support system 1 to identify the child AI 201. For example, the child AI identification data is an ID or name assigned to the child AI 201. The child AI identification data indicates the child AI 201 specialized for a task corresponding to the training input data. In other words, the child AI identification data indicates the child AI 201 that is the correct answer during learning. In the present embodiment, the case where one piece of child AI identification data is associated with one piece of training input data (that is, when there is one piece of child AI identification data that is the correct answer for one piece of training input data) is taken as an example, but a plurality of pieces of child AI identification data may be associated with one piece of training input data.

[0045] For example, when the training input data indicates content related to machine translation, the child AI identification data corresponding to the training input data indicates the child AI 201A specialized for the task of machine translation. When the training input data indicates content related to summary creation, the child AI identification data corresponding to the training input data indicates the child AI 201B specialized for the task of summary creation. When the training input data indicates content related to research, the child AI identification data corresponding to the training input data indicates the child AI 201C specialized for the task of research. When the training input data indicates content related to schedule adjustment, the child AI identification data corresponding to the training input data indicates the child AI 201D specialized for the task of schedule adjustment.

[0046] Note that the training data may be created by a person in charge of the learning of the parent AI 200. For example, the person in charge creates the training input data, which is the input part of the training data, based on existing thread posts or virtual posts. The person in charge determines an appropriate task for these posts and designates (annotates) the child AI identification data of the child AI 201 specialized for the task as the output part of the training data. The learning terminal 10 generates a pair of these input part and output part as training data and stores it in the training database DB. The training data may be generated not by manual designation by the person in charge but by a known tool.

[0047] For example, the data storage unit 1000 stores a plurality of child AIs 201. In the present embodiment, an example is given where the child AI 201 is a transformer-based large language model, similar to the parent AI 200. The child AI 201 may be a large language model using a method other than a transformer. The child AI 201 may be a machine learning model not classified as a large language model. For example, the child AI 201 may be an adversarial generation network, a neural network, or a support vector machine. Further, the child AI 201 may be a program not classified as a machine learning model.

[0048] In the present embodiment, an example is given where an operator providing the business support system 1 to a user creates the child AI 201 by performing retraining of a pre-trained large language model provided by another operator. For example, the child AI 201 includes parameters adjusted by learning and a program indicating processes such as calculation of embedded expressions. The program and parameters of the child AI 201 may be various known programs and parameters. This point is the same as described for the program and parameters of the parent AI 200. The program and parameters of the child AI 201 may be the same as those described as an example of the program and parameters of the parent AI 200.

[0049] For example, the data storage unit 1000 stores training data for retraining a pre-trained large language model. That is, the data storage unit 1000 stores the training data of the child AI 201 separately from the training data of the parent AI 200. Since the training data of the child AI 201 may be known training data, illustration thereof is omitted. The training data of the child AI 201 is prepared according to the task in which the child AI 201 specializes. For example, the training data of the child AI 201 includes input data for training and output data indicating the correct output corresponding to the task of the child AI 201. The output data is determined according to the content of the task.

[0050] For example, the training data of the child AI 201A specialized for the task of machine translation includes training input data indicating the text before translation and output data indicating the text after translation. The training data of the child AI 201B specialized for the task of summary creation includes training input data indicating the text to be summarized and output data indicating the correct summary. The training data of the child AI 201C specialized for the task of research includes training input data indicating the content of the research request and output data indicating the correct content of the research. The training data of the child AI 201D specialized for the task of schedule adjustment includes training input data indicating the schedule to be adjusted and output data indicating the correct adjustment result. Similarly, for other tasks, it is sufficient to prepare training data according to the task.

[0051] Note that the data stored in the data storage unit 1000 is not limited to the examples of this embodiment. The data storage unit 1000 only needs to store data necessary for the learning of at least one of the parent AI 200 and the child AI 201. For example, the data storage unit 1000 may store other data such as a learning program indicating a series of processes during learning. Assume that the calculation formula of the loss function calculated during learning is also shown in the learning program. The data storage unit 1000 may store all of the parent AI 200 before learning (the parent AI 200 with initial parameter values), the parent AI 200 after learning (the parent AI 200 with parameters adjusted by learning), the child AI 201 before learning (the child AI 201 with parameters adjusted by pre-learning), and the child AI 201 after learning (the child AI 201 with parameters adjusted by re-learning).

[0052] [Learning Unit] The learning unit 1001 performs learning of at least one of the parent AI 200 and the child AI 201. For example, the learning unit 1001 performs learning of the parent AI 200 based on each of a plurality of pieces of training data stored in the training database DB. In the present embodiment, a case where the learning unit 1001 performs learning of the parent AI 200 based on an algorithm of supervised learning is taken as an example. The learning algorithm may be a known algorithm. The learning unit 1001 may perform learning of the parent AI 200 based on an algorithm of semi-supervised learning or unsupervised learning.

[0053] In the present embodiment, the learning unit 1001 executes a series of learning processes by executing a learning program stored in the data storage unit 1000. For example, when the input part of the training data stored in the training database DB is input to the parent AI 200, the learning unit 1001 may perform learning of the parent AI 200 by adjusting the parameters of the parent AI 200 so that the output part of the training data is output from the parent AI 200. The learning unit 1001 calculates the loss of the parent AI 200 based on the loss function described in the learning program. The learning unit 1001 repeats the learning until the loss becomes sufficiently small.

[0054] In the present embodiment, since a case where a transformer-based large language model corresponds to the parent AI 200 is taken as an example, the learning unit 1001 may perform learning of the parent AI 200 based on a known algorithm adopted in the transformer-based large language model. The learning unit 1001 may perform learning of the parent AI 200 based on a known algorithm according to the type of the machine learning model adopted as the parent AI 200. For example, the learning unit 1001 may perform learning of the parent AI 200 based on a known algorithm such as the error backpropagation method or the gradient descent method. The learning unit 1001 may perform learning of the parent AI 200 from scratch instead of re-learning a pre-trained large language model. The learning unit 1001 may perform learning of the parent AI 200 based on unannotated training data.

[0055] For example, when the learning in the learning unit 1001 is completed, the learned parent AI 200 is recorded in the data storage unit 1000. The data storage unit 1000 may store both the parent AI 200 before learning and the learned parent AI 200. The learning unit 1001 transmits the learned parent AI 200 to the server 20. The learned parent AI 200 transmitted to the server 20 is provided for user use.

[0056] For example, the learning unit 1001 executes the learning of the child AI 201 based on the training data of the child AI 201. A specific example of the learning method of the child AI 201 may be the same as the method described as an example of the learning method of the parent AI 200. When the input part of the training data of the child AI 201 is input to the child AI 201, the learning unit 1001 executes the learning of the child AI 201 by adjusting the parameters of the child AI 201 so that the output part of the training data is output from the child AI 201. In the present embodiment, the learning unit 1001 executes the relearning of the pre-learned child AI 201. The relearning may be executed by a known algorithm. The learning unit 1001 may execute the learning of the child AI 201 from scratch instead of the relearning of the pre-learned large language model. The learning unit 1001 may execute the learning of the child AI 201 based on the unannotated training data.

[0057] For example, when the learning in the learning unit 1001 is completed, the learned child AI 201 is recorded in the data storage unit 1000. The data storage unit 1000 may store both the child AI 201 before learning and the learned child AI 201. The learning unit 1001 transmits the learned child AI 201 to the server 20. The learned child AI 201 transmitted to the server 20 is provided for user use.

[0058] [3-2. Functions Implemented on the Server] For example, the server 20 includes a data storage unit 2000, an input data acquisition unit 2001, a selection unit 2002, a task processing unit 2003, and a provision unit 2004. The data storage unit 2000 is realized by the storage unit 22. Each of the input data acquisition unit 2001, the selection unit 2002, the task processing unit 2003, and the provision unit 2004 is realized by the control unit 21.

[0059] [Data storage unit] The data storage unit 2000 stores data necessary for business support. For example, the data storage unit 2000 stores the parent AI 200 and a plurality of child AIs 201. In this embodiment, the learning terminal 10 executes the learning of each of the parent AI 200 and the plurality of child AIs 201. The server 20 acquires the learned parent AI 200 and the learned plurality of child AIs 201 from the learning terminal 10 and records them in the data storage unit 2000.

[0060] As described above, since the child AI 201 is an example of a business support program, the part described as the child AI 201 can be read as a business support program. In this embodiment, the case where the child AI 201 corresponds to a certain one specific task is taken as an example, but the child AI 201 may correspond to a plurality of tasks. That is, the child AI 201 may be a single-task AI or a multi-task AI.

[0061] Also, at least one of the parent AI 200 and the child AI 201 may be stored in a computer other than the server 20 (for example, another server computer). For example, when the parent AI 200 is stored in another computer, the selection unit 2002 described later requests the selection of the child AI 201 by transmitting input data to the other computer. The other computer can select the child AI 201 based on the input data received from the server 20 and the parent AI 200 stored in itself. The other computer transmits selection result data indicating the selection result of the child AI 201 to the server 20. The selection unit 2002 may identify the child AI 201 selected by the parent AI 200 based on the selection result data received from the other computer.

[0062] For example, when the child AI 201 is stored in another computer, the task processing unit 2003 described below requests the other computer to process a task by transmitting input data. When the parent AI 200 is not stored in the other computer, the task processing unit 2003 may transmit selection result data indicating the child AI 201 selected by the parent AI 200 to the other computer. Based on these data, the other computer causes the child AI 201 specialized in the task corresponding to the input data to process the task. The other computer transmits processing result data indicating the processing result of the task to the server 20. The task processing unit 2003 acquires the processing result data received from the other computer. In this way, the task processing unit 2003 may process the task by requesting the other computer to substantially process the task and acquiring the processing result data.

[0063] Note that the data stored in the data storage unit 2000 is not limited to the above example. The data storage unit 2000 can store any data. For example, the data storage unit 2000 may store programs and data of each function of the business support system 1 (for example, in the case of a communication function, data of posts to a thread). The data storage unit 2000 may store a database in which various data of users who use the business support system 1 are stored. The data storage unit 2000 may store data such as data for displaying the business support screen SC (for example, HTML data).

[0064] [Input Data Acquisition Unit] The input data acquisition unit 2001 acquires input data indicating the user's input. The user's input can also be referred to as the user's operation. For example, the input data indicates a character string input by the user. The character string can also be referred to as text, a message, or an article. In the example of FIG. 2, the input data indicates the user's post. The input data may indicate other inputs other than a character string. For example, the input data may indicate the voice uttered by the user, the selection result of a file by the user, the selection result of information (e.g., an image or text) displayed on the business support screen SC, or other inputs.

[0065] For example, when the user performs some input on the business support screen SC on the user terminal 30, the user terminal 30 transmits input data indicating the input to the server 20. The input data acquisition unit 2001 acquires the input data from the user terminal 30. The input data acquisition unit 2001 may acquire the input data from the user terminal 30 via a computer other than the server 20 and the user terminal 30. In the example of FIG. 2, when the user performs an operation for posting (e.g., selection of the "Write" button), the user terminal 30 transmits input data indicating the post input in the input form F10 to the server 20. The input data acquisition unit 2001 acquires the input data indicating the user's post from the user terminal 30.

[0066] Note that the input data may be stored in the data storage unit 2000. For example, the input data acquisition unit 2001 records the input data acquired from the user terminal 30 in the data storage unit 2000. The input data acquisition unit 2001 can acquire the input data from the data storage unit 2000 at an arbitrary timing. In this case, the data storage unit 2000 may also store data for identifying the user who made the input indicated by the input data.

[0067] [Selection unit] The selection unit 2002 can select a business support program specialized for a task corresponding to the input data, based on the parent AI 200 that can select a business support program specialized for a specific task in business support from among a plurality of business support programs. In the present embodiment, since the case where the child AI 201 corresponds to a business support program is taken as an example, the selection unit 2002 can select the child AI 201 specialized for a task corresponding to the input data, based on the parent AI 200. The selection unit 2002 may select a plurality of child AIs 201.

[0068] The selection unit 2002 does not necessarily have to select the child AI 201, and it may not select the child AI 201. For example, when there is no task corresponding to the input data, or when there is a task corresponding to the input data but there is no child AI 201 specialized for it, the selection unit 2002 does not have to select the child AI 201. The selection unit 2002 may attempt to select the child AI 201 each time the input data is acquired. When a certain input data is acquired, the selection unit 2002 may not select the child AI 201 specialized for the task corresponding to the input data, and when other input data is acquired, the selection unit 2002 may select the child AI 201 specialized for the task corresponding to the other input data.

[0069] For example, the selection unit 2002 inputs the input data to the parent AI 200. The input data is used as a prompt for the parent AI 200. In addition to the input data, a default prompt prepared in the business support system 1 may be input to the parent AI 200. The parent AI 200 calculates an embedded representation based on the input data, and outputs output data indicating an output corresponding to the embedded representation. When there is a child AI 201 corresponding to the embedded representation, the parent AI 200 outputs the child AI identification data of the child AI 201 as the output data. When there is no child AI 201 corresponding to the embedded representation, the parent AI 200 outputs output data indicating that the child AI 201 was not selected. The selection unit 2002 acquires the output data output from the parent AI 200. A series of processes of the parent AI 200 are executed based on parameters adjusted by learning.

[0070] In this embodiment, taking the case where a transformer-based large language model corresponds to the parent AI 200 as an example, the parent AI 200 divides the character string indicated by the input data into a plurality of tokens, and then calculates the embedding representation of each token. Based on the order of the embedding representations of each token, the parent AI 200 predicts the continuation as necessary and then outputs output data. When there is a child AI 201 corresponding to the order of the embedding representations of each token, the parent AI 200 outputs the child AI identification data of the child AI 201 as output data. When there is no child AI 201 corresponding to the order of the embedding representations of each token, the parent AI 200 outputs output data indicating that the child AI 201 was not selected.

[0071] Note that the processing executed by the parent AI 200 is not limited to the above example. The parent AI 200 may output output data indicating the selection result of the child AI 201 based on the input data. These input and output relationships only need to be defined in the program code of the parent AI 200. For example, when the parent AI 200 is a machine learning model other than a transformer, similarly to the above, the parent AI 200 may calculate the embedding representation based on the input data and output the output data according to the embedding representation. When the parent AI 200 is a program that is not classified as a machine learning model, the output data may be output from the parent AI 200 based on the program code, variables, etc. described in the program.

[0072] [Task Processing Unit] When the task support program is selected by the selection unit 2002, the task processing unit 2003 processes the task corresponding to the input data based on the task support program. In this embodiment, taking the case where the child AI 201 corresponds to the task support program as an example, the task processing unit 2003 processes the task based on the input data and the child AI 201 selected by the selection unit 2002. That is, the task processing unit 2003 causes the child AI 201 selected by the selection unit 2002 to process the task corresponding to the input data.

[0073] For example, the task processing unit 2003 inputs input data to the child AI 201 selected by the selection unit 2002. The input data is used as a prompt for the child AI 201. In addition to the input data, a default prompt prepared in the business support system 1 may also be input to the child AI 201. Based on the input data, the child AI 201 calculates an embedding representation and outputs output data corresponding to the embedding representation. The child AI 201 outputs output data corresponding to the embedding representation. The output of the output data corresponds to the processing of a task according to the input data. The selection unit 2002 acquires the output data output from the child AI 201. A series of processes of the child AI 201 are executed based on parameters adjusted by learning.

[0074] For example, when the child AI 201A specialized in the task of machine translation is selected by the selection unit 2002, the child AI 201A outputs output data indicating a translation result corresponding to the embedding representation. When the child AI 201B specialized in the task of summary creation is selected by the selection unit 2002, the child AI 201B outputs output data indicating a summary corresponding to the embedding representation. When the child AI 201C specialized in the task of research is selected by the selection unit 2002, the child AI 201C outputs output data indicating information corresponding to the embedding representation. When the child AI 201D specialized in the task of schedule adjustment is selected by the selection unit 2002, the child AI 201D outputs output data indicating an adjustment result corresponding to the embedding representation.

[0075] In this embodiment, taking the case where a transformer-based large language model corresponds to the child AI 201 as an example, the child AI 201 divides the character string indicated by the input data into a plurality of tokens and then calculates the embedding representation of each token. Based on the order of the embedding representations of each token, the child AI 201 predicts the continuation if necessary and then outputs the output data.

[0076] Note that the processing executed by the child AI 201 is not limited to the above example. The child AI 201 may output output data indicating the selection result of the child AI 201 based on the input data. These input and output relationships only need to be defined in the program code of the child AI 201. For example, even when the child AI 201 is a machine learning model other than a transformer, similarly to the above, the child AI 201 may calculate an embedding representation based on the input data and output output data corresponding to the embedding representation. When the child AI 201 is a program that is not classified as a machine learning model, output data may be output from the child AI 201 based on the program code, variables, etc. described in the program.

[0077] In this embodiment, when the child AI 201 is not selected by the selection unit 2002, the task processing unit 2003 does not process the task according to the input data. In this case, the task processing unit 2003 transmits data indicating that the task corresponding to the input data has not been processed to the user terminal 30. The user terminal 30 causes the display unit 35 to display an error message indicating that the task corresponding to the input data has not been processed based on the data. When the child AI 201 is not selected by the selection unit 2002, the task processing unit 2003 may cause a general-purpose parent AI 200 to process the task as in a modified example described later. For example, when the child AI 201 is not selected by the selection unit 2002, the task processing unit 2003 may cause a general-purpose AI prepared separately from the parent AI 200 to process the task.

[0078] [Provision unit] The provision unit 2004 provides the processing result of the task processing unit 2003 to the user. In this embodiment, since the provision unit 2004 is realized by the server 20, the provision unit 2004 provides the processing result to the user by transmitting processing result data indicating the processing result of the task processing unit 2003 to the user terminal 30. The user terminal 30 causes the business support screen SC to display information (for example, an image or text) indicating the processing result of the task processing unit 2003 based on the processing result data.

[0079] For example, the providing unit 2004 includes an output providing unit 2004A. The output providing unit 2004A provides the user with the output from the business support program selected by the selection unit 2002 and acquired by the task processing unit 2003. In this embodiment, since the child AI 201 corresponds to the business support program, the output providing unit 2004A provides the user with the output from the child AI 201 acquired by the task processing unit 2003. That is, the output providing unit 2004A provides the output acquired by the task processing unit 2003 to the user as it is. The output data output from the child AI 201 is an example of the processing result data. Instead of the output data directly becoming the processing result data, the processing result data may be generated based on the output data as in the modification example described later.

[0080] Note that the processing result data may be in any format. In this embodiment, the processing result data may be data necessary for displaying some information on the business support screen SC. For example, when the business support screen SC is displayed on a browser, the processing result data may be data in a markup language such as HTML. When the business support screen SC is displayed by an application dedicated to the business support system 1, the processing result data may be data (for example, image data or text data) necessary for displaying some information in the application. In the example of FIG. 2, the providing unit 2004 causes the answer indicated by the processing result data to be displayed as a reply to the user's post. The user's post is also displayed on the business support screen SC by the processing of the server 20.

[0081] [3-3. Functions Implemented on the User Terminal] For example, the user terminal 30 includes a data storage unit 3000, a display control unit 3001, and an input reception unit 3002. The data storage unit 3000 is realized by the storage unit 32. Each of the display control unit 3001 and the input reception unit 3002 is realized by the control unit 31.

[0082] [Data Storage Unit] The data storage unit 3000 stores data for business support. For example, the data storage unit 3000 stores browsers for displaying various screens of the business support system 1. For example, the data storage unit 3000 stores programs dedicated to the business support system 1. The business support screen SC may be displayed on a program dedicated to the business support system 1.

[0083] [Display control unit] The display control unit 3001 causes various screens in the business support system 1 to be displayed on the display unit 35. For example, the display control unit 3001 causes various screens such as the business support screen SC to be displayed on the display unit 35 based on the data received from the server 20.

[0084] [Input reception unit] The input reception unit 3002 receives user input. For example, the input reception unit 3002 receives input for various screens such as the business support screen SC. Input data indicating the input received by the input reception unit 3002 is appropriately transmitted to the server 20.

[0085] [4. Processes executed in the business support system] FIG. 6 is a diagram showing an example of a process executed in the business support system 1. The control units 11, 21, 31 execute the programs stored in the storage units 12, 22, 32, respectively, whereby the process of FIG. 6 is executed. The process of FIG. 6 is an example of a process included in the business support method.

[0086] For example, the learning terminal 10 executes learning of the parent AI 200 based on the training data of the parent AI 200 stored in the training database DB (S1). For each pre-trained child AI 201, the learning terminal 10 executes learning of the child AI 201 based on the training data of the child AI 201 (S2). The learning terminal 10 transmits the learned parent AI 200 and the plurality of learned child AIs 201 to the server 20 (S3). The server 20 acquires the parent AI 200 and the plurality of child AIs 201 from the learning terminal 10 (S4). Thereafter, the user can use the parent AI 200 and the plurality of child AIs 201.

[0087] The user terminal 30 executes a login process for the user to log in to the business support system 1 with the server 20 (S5). The user terminal 30 receives the user's input based on the detection signal of the operation unit 34 (S6). The user terminal 30 transmits input data indicating the input received in S to the server 20 (S7). The server 20 acquires the input data from the user terminal 30 (S8). In the example of FIG. 2, it is assumed that subsequent processing is executed when an AI such as the parent AI 200 and the child AI 201 is mentioned.

[0088] The server 20 inputs the input data to the parent AI 200 (S9). The server 20 acquires output data indicating the output from the parent AI 200 (S10). The server 20 determines whether the child AI 201 selected by the parent AI 200 is shown in the output data (S11). In S11, if it is determined that the child AI 201 is not shown in the output data (S11: N), this process ends. In this case, a message indicating that the task processing by the child AI 201 was not performed may be displayed on the user terminal 30.

[0089] In S11, if it is determined that the child AI 201 is shown in the output data (S11: Y), the server 20 causes the child AI 201 to process a task according to the input data (S12). The server 20 acquires output data indicating the output from the child AI 201 (S13). The server 20 executes a process for providing the output from the child AI 201 to the user to the user terminal 30 (S14), and this process ends. In the example of FIG. 2, the answer from the child AI 201 is provided to the user as an output by the process of S14.

[0090] [Summary of Embodiment] The business support system 1 of this embodiment can select a child AI 201 specialized in a task according to input data based on the parent AI 200. When the child AI 201 is selected, the business support system 1 processes a task according to the input data based on the child AI 201. Thereby, since the business support system 1 can effectively support the user's business, the convenience of the user can be enhanced. For example, since the user does not need to select the child AI 201 for processing the task from among a plurality of child AIs 201, the business support system 1 can save the user's labor. If the user selects the child AI 201 by himself / herself, the user may select a child AI 201 that is not appropriate for processing the task according to the input data. However, the business support system 1 can select an appropriate child AI 201 by the parent AI 200 and process the task appropriately.

[0091] In addition, the business support system 1 provides the output from the child AI 201 selected by the selection unit 2002 to the user. By providing the output from the child AI 201 to the user as the processing result of the task according to the input data, the business support system 1 can effectively support the user's business. The user can check the content of the output from the child AI 201 as it is.

[0092] [6. Modification Example] Note that the present disclosure is not limited to the embodiments described above. The present disclosure can be appropriately changed without departing from the gist of the present disclosure.

[0093] [6-1. Modification Example 1] For example, in the embodiment, the case where the output providing unit 2004A provides the output from the child AI 201 specialized in the task according to the input data to the user as it is was taken as an example. In Modification Example 1, the case where the parent AI 200 processes the output from the child AI 201 in an appropriate form is taken as an example. For example, the parent AI 200 corrects the output from the child AI 201 into general words. The parent AI 200 may merge the output from the child AI 201 and the processing result of the task it has processed. In Modification Example 1, the output after being processed by the parent AI 200 is provided to the user.

[0094] FIG. 7 is a diagram showing an example of functions realized by the business support system 1 of Modification 1. The business support system 1 of Modification 1 includes a processing execution unit 2005. The processing execution unit 2005 is realized by the control unit 11. The providing unit 2004 of Modification 1 includes a processing result providing unit 2004B. The providing unit 2004 of Modification 1 may not include the output providing unit 2004A. The business support system 1 may include both the output providing unit 2004A and the processing result providing unit 2004B, and these may be used appropriately according to the situation.

[0095] The processing execution unit 2005 executes processing on the output from the business support program selected by the selection unit 2002 and acquired by the task processing unit 2003 based on the parent AI 200. Also in Modification 2, as in the embodiment, the case where the child AI 201 corresponds to the business support program is taken as an example. For example, the processing execution unit 2005 inputs the output data from the child AI 201 to the parent AI 200. The parent AI 200 may modify all or part of the output indicated by the output data from the child AI 201, or may add all or part of the content of the input indicated by the input data to the output indicated by the output data from the child AI 201.

[0096] Note that the processing execution unit 2005 may input to the parent AI 200 at least one of not only the output data from the child AI 201 but also the input data and the prompt data indicating a predetermined prompt. The processing execution unit 2005 may execute processing on the output indicated by the output data based at least on the output data from the child AI 201. The processing content for the output from the child AI 201 may be indicated in the prompt data.

[0097] For example, when the parent AI 200 is to modify the output from the child AI 201 into general terms, the prompt data shows a prompt such as "Please modify this text into general terms." When the parent AI 200 is to merge the output from the child AI 201 and its own processing result, the prompt data shows a prompt such as "Please merge this text and your processing result." The parent AI 200 may execute the processing of the processing content indicated by the prompt data on the output from the child AI.

[0098] The processing result providing unit 2004B provides the processing result of the processing execution unit 2005 to the user. The processing result providing unit 2004B provides the processing result to the user by transmitting processing result data indicating the processing result of the processing execution unit 2005 to the user terminal 30. The user terminal 30 displays information (for example, an image or text) indicating the processing result of the processing execution unit 2005 on the business support screen SC based on the processing result data. The processing result providing unit 2004B is different from the output providing unit 2004A in that it does not directly provide the output from the child AI 201 to the user, but the information providing method itself is the same as that of the output providing unit 2004A.

[0099] The business support system 1 of Modification Example 1 executes processing on the output from the business support program selected by the selection unit 2002 and acquired by the task processing unit 2003 based on the parent AI 200. The business support system 1 provides the processing result by the parent AI 200 to the user. Thereby, the business support system 1 can provide more appropriate information to the user, so that the convenience of the user can be effectively improved. For example, when the output from the child AI 201 is unnatural as a text, the business support system 1 can modify it into an appropriate text by the parent AI 200 and then provide the modified text to the user, thus improving the convenience of the user. When the business support system 1 merges the output from the child AI 201 and the processing result of the task by the parent AI 200 and provides it to the user, more information can be presented to the user.

[0100] [6-2. Modification Example 2] For example, when the task processing unit 2003 does not have a business support program selected by the selection unit 2002, it may process a task according to the input data based on the parent AI 200. In Modification Example 2, as in the embodiment, the case where the child AI 201 corresponds to a business support program is taken as an example. Therefore, when the child AI 201 is not selected by the selection unit 2002, the task processing unit 2003 processes a task according to the input data based on the parent AI 200.

[0101] The parent AI 200 in Modification Example 2 is not an AI specialized for a specific task, but a general-purpose AI capable of processing various tasks. For example, in the embodiment, the case where the parent AI 200 is created by re-training a pre-trained large language model is taken as an example. However, when the pre-trained large language model is a transformer-based model such as GPT, the parent AI 200 created by pre-training also has versatility. The parent AI 200 may be a multi-task AI capable of handling a plurality of specific tasks.

[0102] For example, the task processing unit 2003 inputs input data to the parent AI 200 selected by the selection unit 2002. The parent AI 200 calculates an embedding representation based on the input data and outputs output data according to the embedding representation. The parent AI 200 outputs output data according to the embedding representation. The output of the output data corresponds to the processing of a task according to the input data. The selection unit 2002 acquires the output data output from the parent AI 200. A series of processes of the parent AI 200 may be executed based on parameters adjusted by learning. For example, when re-training for imparting versatility to the parent AI 200 is executed, a series of processes of the parent AI 200 are executed based on the parameters adjusted by re-training.

[0103] When the child AI 201 is not selected by the selection unit 2002, the business support system 1 of Modification Example 2 processes a task according to the input data based on the parent AI 200. Thereby, when the business support program is not selected by the selection unit 2002, the business support system 1 can prevent the situation where no information can be provided to the user and can provide some information to the user. For example, when a task that cannot be handled by the child AI 201 is requested by the user, the business support system 1 can process the task based on the general-purpose parent AI 200 and provide the processing result of the user.

[0104] [6-3. Modification Example 3] For example, in the embodiment, although the case where each of the plurality of business support programs is the child AI 201 has been described, the plurality of business support programs may include a child AI 201 which is an AI different from the parent AI 200 and a non-AI program which is a non-AI. The non-AI program is a program that is not classified as an AI. All or part of the input data is input to the non-AI program. The non-AI program outputs output data based on all or part of the input data input to itself. The non-AI program is assumed to be stored in the data storage unit 2000, but the non-AI program may be stored in another computer. The non-AI program may be a simple program such as a program for acquiring the current date. Even if it is such a simple program, it can process the task of acquiring the current date.

[0105] For example, the non-AI program may be a machine translation program not classified as an AI, a summary creation program not classified as an AI, a research program not classified as an AI (for example, a search engine program not classified as an AI), a schedule adjustment program not classified as an AI, or other programs. These non-AI programs may be known groupware or programs employed in known business support services not classified as groupware. The non-AI program may be the rule-based program described above.

[0106] When the task processing unit 2003 of Modification 3 selects the child AI 201 by the selection unit 2002, it processes a task according to the input data based on the child AI 201. The processing of the task processing unit 2003 in this case is as described in the embodiment. When the non-AI program is selected by the selection unit 2002, the task processing unit 2003 processes a task according to the input data based on the non-AI program. The task processing unit 2003 inputs all or part of the input data to the non-AI program. The processing of the non-AI program may be processing adopted in known groupware or known services for business support that are not classified as groupware.

[0107] For example, the non-AI program executes processing on all or part of the input data input to itself based on the program code included in itself. The non-AI program outputs output data indicating the execution result of the program code. The task processing unit 2003 acquires the output data output from the non-AI program. When the non-AI program is stored in another computer other than the server 20, the task processing unit 2003 may transmit the input data to the other computer and request the processing of the task by the non-AI program. The task processing unit 2003 acquires the processing result data indicating the processing result of the task by the non-AI program from the other computer.

[0108] When the business support system 1 of Modification 3 selects the child AI 201 by the selection unit 2002, it processes a task according to the input data based on the child AI 201. When the non-AI program is selected by the selection unit 2002, the business support system 1 processes a task according to the input data based on the non-AI program. Thereby, the business support system 1 can cause the appropriate one of the child AI 201 and the non-AI program to process a task according to the input data.

[0109] [6-4. Modification 4] For example, in Modification 3, there may be child AIs 201 and non-AI programs specialized for the same task. For example, not only the child AI 201A specialized for machine translation, but also a non-AI program not classified as an AI may be able to process the task of machine translation. Not only the child AI 201B specialized for summary creation, but also a non-AI program not classified as an AI may be able to process the task of summary creation. Not only the child AI 201C specialized for research, but also a non-AI program not classified as an AI may be able to process the task of research. Not only the child AI 201D specialized for schedule adjustment, but also a non-AI program not classified as an AI may be able to process the task of schedule adjustment.

[0110] The selection unit 2002 of Modification 4 selects both the child AI 201 and the non-AI program when both are capable of processing the task according to the input data. When both the child AI 201 and the non-AI program are selected by the selection unit 2002, the task processing unit 2003 processes the task according to the input data based on both the child AI 201 and the non-AI program, and may select either the output from the child AI 201 or the output from the non-AI program. The processing of the task by each of the child AI 201 and the non-AI program is as described in the embodiment or Modification 3.

[0111] The task processing unit 2003 selects one of the output from the child AI 201 and the output from the non-AI program based on a predetermined selection method. The selection method may be any predetermined method. In Modification 4, as an example of the selection method, a case where the parent AI 200 is made to select an appropriate one of the output from the child AI 201 and the output from the non-AI program will be given as an example. For example, in the training data of the parent AI 200, the relationship between the two training outputs and the correct output is shown. The learning unit 1001 executes the learning of the parent AI 200 so that when the two training outputs indicated by the input part of the training data are input to the parent AI 200, the parent AI 200 selects and outputs the correct output indicated by the output part of the training data. The learning algorithm may be various known algorithms as in the embodiment.

[0112] For example, the task processing unit 2003 inputs each of the output from the child AI 201 and the output from the non-AI program to the parent AI 200. The parent AI 200 calculates the embedding representation of each of the output from the child AI 201 and the output from the non-AI program, and selects an appropriate one based on these embedding representations. Depending on the embedding representations of the output from the child AI 201 and the output from the non-AI program, the parent AI 200 may select both of them or only one of them. The parent AI 200 outputs output data indicating the selection result of at least one of the output from the child AI 201 and the output from the non-AI program.

[0113] For example, the task processing unit 2003 acquires the output data output from the parent AI 200. The providing unit 2004 provides the processing result to the user based on the selection result indicated by the output data. For example, the providing unit 2004 provides the processing result to the user based on at least one selected by the parent AI 200 among the output from the child AI 201 and the output from the non-AI program.

[0114] Note that the selection method of the task processing unit 2003 is not limited to the method used by the parent AI 200. The task processing unit 2003 may select one of the output from the child AI 201 and the output from the non-AI program based on other selection methods. For example, when the child AI 201 can calculate a score indicating the accuracy of the output and the non-AI program can also calculate a score indicating the accuracy of the output, the task processing unit 2003 may select the one with the higher score between the output from the child AI 201 and the output from the non-AI program. Another example is that the task processing unit 2003 may select the one with the longer text between the output from the child AI 201 and the output from the non-AI program.

[0115] In the business support system 1 of Modification Example 4, when both the child AI 201 and the non-AI program are selected by the selection unit 2002, the task corresponding to the input data is processed based on both the child AI 201 and the non-AI program, and one of the output from the child AI 201 and the output from the non-AI program is selected. Thereby, the business support system 1 can provide a more appropriate processing result to the user.

[0116] [6-5. Modification Example 5] For example, as a plurality of business support programs, it may include a public program that processes publicly available data and a non-public program that processes non-public data that cannot be made public. In Modification Example 5, similar to the embodiment, the case where the child AI 201 corresponds to a business support program is taken as an example. The child AI 201 that processes publicly available data is an example of a public program. The child AI 201 that processes non-public data is an example of a non-public program. The child AI 201 that processes publicly available data may exist in another computer other than the server 20. The information processing of the child AI 201 that processes non-public data is completed in a secure environment within the server 20.

[0117] The public data is data that may be publicly disclosed outside the business support system 1. For example, data that may be left as a log outside the business support system 1 corresponds to the public data. The public data is a part of the data registered in the business support system 1. For example, data other than the confidential data of the organization to which the user belongs corresponds to the public data. Data indicating information posted on a general website such as an Internet encyclopedia or a homepage corresponds to the public data. Classification data indicating which data is classified as public data is assumed to be pre-stored in the data storage unit 2000.

[0118] The non-public data is data that must not be publicly disclosed outside the business support system 1. For example, data that must not be left as a log outside the business support system 1 corresponds to the non-public data. The non-public data is data other than the public data among the data registered in the business support system 1. For example, among the data of the schedule registered in the schedule management function, data such as the names of people outside the organization to which the user belongs corresponds to the non-public data. Another example is that the confidential data of the organization to which the user belongs corresponds to the non-public data. Classification data indicating which data is classified as non-public data is assumed to be pre-stored in the data storage unit 2000. The child AI 201 of Modification Example 5 refers to at least one of the public data and the non-public data and processes the task.

[0119] In Modification Example 5, the output part of the training data learned by the parent AI 200 is child AI identification data indicating either the child AI 201 that processes the public data or the child AI 201 that processes the non-public data. For example, for the processing of a task according to a certain training input data, if only the public disclosure of the public data is required and the public disclosure of the non-public data is not required, the child AI identification data corresponding to the pair of the training input data indicates the child AI 201 that processes the public data. For example, for the processing of a task according to a certain training input data, if not only the public disclosure of the public data but also the public disclosure of the non-public data is required, the child AI identification data corresponding to the pair of the training input data indicates the child AI 201 that processes the non-public data.

[0120] Since the parent AI 200 of Modification 5 has learned higher-level training data, it can estimate which child AI 201 for processing public data or which child AI 201 for processing non-public data should be selected for any input data. For example, the selection unit 2002 inputs input data to the parent AI 200. The parent AI 200 calculates the embedded representation of the input data and outputs output data indicating a child AI 201 for processing public data or a child AI 201 for processing non-public data based on the embedded representation. As described in the embodiment, the parent AI 200 may not select any child AI 201.

[0121] When the public program is selected by the selection unit 2002, the task processing unit 2003 of Modification 5 processes a task according to the input data based on the public program and the public data. In Modification 5, taking the case where the child AI 201 for processing public data corresponds to the public program as an example, the task processing unit 2003 inputs the input data to the child AI 201 for processing public data. The child AI 201 calculates the embedded representation of the input data and acquires public data based on the embedded representation. Instead of the child AI 201 acquiring the public data, the parent AI 200 may acquire the public data and the selection unit 2002 may input the public data to the child AI 201. The child AI 201 processes a task according to the input data after leaving the public data as a log outside the business support system 1. Although it is different from the embodiment and other modifications in that the public data is left as a log outside the business support system 1, the flow of the child AI 201 processing the task is the same as that of the embodiment and other modifications.

[0122] When the non-public program is selected by the selection unit 2002 in the task processing unit 2003 of Modification Example 5, a task corresponding to the input data is processed based on the non-public program and the non-public data. In Modification Example 5, since the case where the child AI 201 that processes the non-public data corresponds to the non-public program is taken as an example, the task processing unit 2003 inputs the input data to the child AI 201 that processes the non-public data. The child AI 201 calculates the embedded representation of the input data and acquires the non-public data based on the embedded representation. Instead of the child AI 201 acquiring the non-public data, the parent AI 200 may acquire the non-public data and the selection unit 2002 may input the non-public data to the child AI 201. The child AI 201 processes the task corresponding to the input data without leaving the non-public data as a log outside the business support system 1. Although it is different from the embodiment and other modification examples in that the non-public data is not left as a log outside the business support system 1, the flow of the child AI 201 processing the task is the same as that of the embodiment and other modification examples.

[0123] When the public program is selected by the selection unit 2002 in the business support system 1 of Modification Example 5, a task corresponding to the input data is processed based on the public program and the public data. When the non-public program is selected by the selection unit 2002 in the business support system 1, a task corresponding to the input data is processed based on the non-public program and the non-public data. Thereby, the business support system 1 can surely prevent the confidential data from leaking to the outside. For example, the business support system 1 can selectively use the child AI 201 that processes the non-public data and the child AI 201 that processes the public data as the child AI 201 capable of processing the same task.

[0124] [6-6. Modification Example 6] For example, when there are many child AIs 201 specialized for a specific task, a single parent AI 200 may not be able to select an appropriate child AI 201. Therefore, there may be three or more hierarchies in the AI, such as parent, child, and grandchild. In Modification Example 6, a three - layer hierarchy is taken as an example, but the number of hierarchies may be four or more. The AI in the lowest hierarchy is an AI specialized for a specific task. The AIs other than those in the lowest hierarchy are AIs that can select an AI in the hierarchy one level below themselves. The AIs other than those in the lowest hierarchy may be general - purpose AIs described in Modification Example 2.

[0125] FIG. 8 is a diagram showing an example of the AI of Modification Example 6. In FIG. 8, there are a parent AI 200, a plurality of child AIs 201, and a plurality of grandchild AIs 202. The parent AI 200 can select at least one child AI 201 based on the input data. Similar to the embodiment, the training data of the parent AI 200 shows the relationship between the input data for training and the child AI identification data of the correct child AI 201. However, the child AI 201 of Modification Example 6 is different from the embodiment in that it is not an AI specialized for a specific task but an AI that can select a grandchild AI 202. Since such training data has been learned by the parent AI 200 of Modification Example 6, it can select an appropriate child AI 201 according to the input data.

[0126] The child AI 201 of Modification Example 6 can select at least one grandchild AI 202 based on the input data. In Modification Example 6, the child AI 201 has the same functions as the parent AI 200 described in the embodiment and Modification Examples 1 - 5. The training data of the child AI 201 is the same as the training data of the parent AI 200 described in the embodiment. However, the output part of the training data is the grandchild AI identification data for identifying the grandchild AI 202. The grandchild AI identification data is different from the child AI identification data in that it is data for identifying the grandchild AI 202, but is the same as the child AI identification data in other respects. Since such training data has been learned by the child AI 201 of Modification Example 6, it can select an appropriate grandchild AI 202 according to the input data.

[0127] For example, the grandchild AI 202 has the same functions as the child AI 201 described in the embodiments and modification examples 1 to 5. The description of the child AI 201 in the embodiments and modification examples 1 to 5 can be read as the grandchild AI 202. In the example of FIG. 8, there are a plurality of grandchild AIs 202 specialized in the same task. For example, as the grandchild AI 202 specialized in the task of machine translation, there are a grandchild AI 202A specialized in English translation and a grandchild AI 202B specialized in Chinese translation. The grandchild AI 202A has learned English training data. The grandchild AI 202B has learned Chinese training data. The child AI 201A can estimate which of the grandchild AIs 202A and 202B is appropriate based on the input data and select the appropriate one. Note that the parent AI 200 estimates that the processing of the child AI 201A is appropriate based on the input data and selects the child AI 201A. The child AI 201A has learned the input data for training to be translated as training data.

[0128] For example, as the grandchild AI 202 specialized in the task of summary creation, there are a grandchild AI 202C specialized in creating a summary of a thread and a grandchild AI 202D specialized in creating a summary of an email. The grandchild AI 202C has learned training data indicating thread posts. The grandchild AI 202D has learned training data indicating the text of an email. The child AI 201B can estimate which of the grandchild AIs 202C and 202D is appropriate based on the input data and select the appropriate one. Note that the parent AI 200 estimates that the processing of the child AI 201B is appropriate based on the input data and selects the child AI 201B. The child AI 201B has learned the input data for training to be summarized as training data. Similarly for other tasks, the grandchild AI 202 that performs the final task processing is selected by the child AI 201.

[0129] The selection unit 2002 of Modification Example 6 selects the child AI 201 that can select the business support program. The selection unit 2002 selects the business support program based on the selected child AI 201. In the example of FIG. 8, the selection unit 2002 may correspond to the parent AI 200. When the selection unit 2002 does not correspond to the parent AI 200, the selection unit 2002 may select the child AI 201 based on the parent AI 200. For example, the selection unit 2002 selects the business support program based on the child AI 201 selected by the parent AI 200, which is another AI capable of selecting any child AI 201 among the plurality of child AIs 201. Also in Modification Example 6, the case where the grandchild AI 202 corresponding to the child AI 201 described in the embodiment corresponds to the business support program is described. At least one of the parent AI 200 and the child AI 201 in Modification Example 6 may be a general-purpose AI. Note that the selection of the child AI 201 may be performed by a program that is not an AI such as the parent AI 200 and is not classified as an AI.

[0130] For example, the selection unit 2002 inputs input data to the parent AI 200. The parent AI 200 calculates the embedded representation based on the input data and outputs output data corresponding to the embedded representation. When there is a child AI 201 corresponding to the embedded representation, the parent AI 200 outputs the child AI identification data of the child AI 201 as the output data. When there is no child AI 201 corresponding to the embedded representation, the parent AI 200 outputs output data indicating that the child AI 201 has not been selected. The selection unit 2002 acquires the output data output from the parent AI 200. The series of processes of the parent AI 200 are executed based on parameters adjusted by learning.

[0131] For example, the selection unit 2002 inputs input data to the child AI 201 indicated by the output data. The parent AI 200 calculates an embedded representation based on the input data and outputs output data corresponding to the embedded representation. If there is a grandchild AI 202 corresponding to the embedded representation, the parent AI 200 outputs the child AI identification data of the grandchild AI 202 as the output data. If there is no grandchild AI 202 corresponding to the embedded representation, the parent AI 200 outputs output data indicating that the grandchild AI 202 was not selected. The selection unit 2002 acquires the output data output from the parent AI 200. A series of processes of the child AI 201 are executed based on parameters adjusted by learning.

[0132] Note that the processing of the task processing unit 2003 after the grandchild AI 202 is selected may be the same as in the embodiment. The task processing unit 2003 may cause the grandchild AI 202 to process what was described as the task to be processed by the child AI 201 in the embodiment. The task processing unit 2003 inputs input data to the grandchild AI 202 selected by the selection unit 2002. The task processing unit 2003 acquires the output data from the grandchild AI 202.

[0133] The business support system 1 of Modification Example 6 selects a child AI 201 that can select a grandchild AI 202. Thereby, even if the number of business support programs including the grandchild AI 202 increases, the business support system 1 can appropriately process tasks according to the input data.

[0134] [6-7. Modification Example 7] For example, when a plurality of business support programs are selected by the selection unit 2002, the task processing unit 2003 may process a task according to the input data based on the plurality of business support programs and merge the outputs from each of the plurality of business support programs. Also in Modification 7, similar to the embodiment, the case where the child AI 201 corresponds to a business support program is taken as an example. When Modification 6 and Modification 7 are combined, the grandchild AI 202 corresponds to a business support program. The merging of the outputs may mean that the task processing unit 2003 simply combines a plurality of outputs, or may mean that the task processing unit 2003 generates a new output based on a plurality of outputs.

[0135] For example, when a plurality of child AIs 201 are selected by the selection unit 2002, the task processing unit 2003 causes each of the plurality of child AIs 201 to process a task according to the input data. The processing of the task by each individual child AI 201 is as described in the embodiment. The task processing unit 2003 combines the outputs from each of the plurality of child AIs 201 to generate one output. For example, the task processing unit 2003 may perform merging by taking the AND of the outputs from each of the plurality of child AIs 201, or may perform merging by taking the OR of the outputs from each of the plurality of child AIs 201. The providing unit 2004 provides the processing result to the user based on the one output data generated by the task processing unit 2003.

[0136] For example, assume that the task according to the input data is schedule adjustment. Further, assume that the user's schedule is separately managed by a plurality of schedule management functions. The task processing unit 2003 merges the output from the child AI 201 that performs schedule adjustment based on the schedule managed by a certain schedule management function and the output from the child AI 201 that performs schedule adjustment based on the schedule managed by another schedule management function. The task processing unit 2003 can obtain an output indicating the time when the user is surely free by merging them so as to take the AND of the free time of the user indicated by these outputs. Similarly for other tasks, merging may be performed according to the child AI 201.

[0137] When a plurality of child AIs 201 are selected by the selection unit 2002, the business support system 1 of Modification Example 7 processes the task according to the input data based on the plurality of child AIs 201 and merges the outputs from each of the plurality of child AIs 201. Thereby, since the business support system 1 can merge the processing results from more child AIs 201, it can provide useful information to the user. For example, even if the user's schedule is separately managed by a plurality of schedule management functions, the business support system 1 can check the free status of the user managed by these schedule management functions for each of the plurality of child AIs 201 and merge the outputs, thereby presenting to the user the time when the user is surely free.

[0138] [6-8. Modification Example 8] For example, when a plurality of business support programs are selected by the selection unit 2002, the task processing unit 2003 may process the task according to the input data by inputting the output from some of the plurality of business support programs to the other business support programs among the plurality of business support programs. Also in Modification Example 8, as in the embodiment, the case where the child AI 201 corresponds to the business support program is taken as an example. When Modification Example 6 and Modification Example 8 are combined, the grandchild AI 202 corresponds to the business support program.

[0139] For example, the task processing unit 2003 may input the output data from the child AI 201A specialized for the machine translation task to the child AI 201B specialized for the task of creating a summary. In this case, the child AI 201B creates a summary in a language other than the input language indicated by the input data. The task processing unit 2003 acquires the output data output by the child AI 201B. For another example, the task processing unit 2003 may input the output data from the child AI 201B specialized for the task of creating a summary to the child AI 201C specialized for the task of research. In this case, the child AI 201C executes research according to the summary of the input indicated by the input data. The task processing unit 2003 acquires the output data output by the child AI 201C.

[0140] Note that regarding which output from which child AI 201 should be input to which child AI 201, it may be determined in advance, or it may be determined by the parent AI 200. When the determination is made by the parent AI 200, the training data of the parent AI 200 includes training data regarding which output from which child AI 201 should be input to which child AI 201. By such training data being learned by the parent AI 200 , the parent AI 200 can estimate which output from which child AI 201 should be input to which child AI 201. The task processing unit 2003 may perform input and output in the flow estimated by the parent AI 200 to obtain the final output.

[0141] In the business support system 1 of Modification Example 8, when a plurality of child AIs 201 are selected by the selection unit 2002, the output from some of the plurality of child AIs 201 among the plurality of child AIs 201 is input to the other child AIs 201 among the plurality of child AIs 201, thereby processing a task according to the input data. Thereby, since the business support system 1 can cooperate a plurality of child AIs 201, the processing accuracy of the task can be improved.

[0142] [6-9. Other Modification Examples] For example, two or more of Modifications 1 to 8 may be combined. For example, when a user uses a communication function such as a thread, the case where AI-based business support is performed has been cited as an example. However, the business support system 1 may perform AI-based business support when the user uses other business support functions. Also in this case, the business support system 1 may perform business support by executing the same processing as in the embodiments and Modifications 1 to 8.

[0143] For example, the functions described as being realized by the server 20 may be realized by the user terminal 30. In this case, the function may be realized by a browser script or an application installed in the user terminal 30. For example, each function may be shared by a plurality of computers or may be realized by one computer.

Explanation of Reference Numerals

[0144] 1 Business support system, 10 Learning terminal, 11, 21, 31 Control unit, 12, 22, 32 Storage unit, 13, 23, 33 Communication unit, 14, 34 Operation unit, 15, 35 Display unit, 20 Server, 30 User terminal, N Network, DB Training database, SC Business support screen, F10 Input form, 1000 Data storage unit, 1001 Learning unit, 2000 Data storage unit, 2001 Input data acquisition unit, 2002 Selection unit, 2003 Task processing unit, 2004 Provision unit, 2004A Output provision unit, 2004B Processing result provision unit, 2005 Processing execution unit, 3000 Data storage unit, 3001 Display control unit, 3002 Input reception unit, 200 Parent AI, 201, 201A, 201B, 201C, 201D Child AI, 202 Grandchild AI.

Claims

1. An input data acquisition unit that acquires input data indicating a user's input, A selection unit capable of selecting the business support program specialized for the task corresponding to the input data from among a plurality of business support programs specialized for specific tasks in business support, based on AI (Artificial Intelligence), A task processing unit that processes the task corresponding to the input data based on the business support program when the business support program is selected by the selection unit, A business support system including the above.

2. The business support system further includes an output providing unit that provides the output obtained by the task processing unit, which is the output from the business support program selected by the selection unit, to the user, The business support system according to Claim 1.

3. The business support system A processing execution unit that executes processing on the output obtained by the task processing unit, which is the output from the business support program selected by the selection unit, based on the AI, A processing result providing unit that provides the processing result of the processing execution unit to the user, The business support system according to Claim 1 or 2, further including the above.

4. When the business support program is not selected by the selection unit, the task processing unit processes the task corresponding to the input data based on the AI, The business support system according to Claim 1 or 2.

5. The plurality of business support programs include an AI different from the above AI and a non-AI program that is non-AI, The task processing unit When the different AI is selected by the selection unit, processes the task corresponding to the input data based on the different AI, When the non-AI program is selected by the selection unit, processes the task corresponding to the input data based on the non-AI program, The business support system according to Claim 1 or 2.

6. When both the different AI and the non-AI program are selected by the selection unit, the task processing unit processes the task corresponding to the input data based on both the different AI and the non-AI program, and selects either the output from the different AI or the output from the non-AI program, The business support system according to claim 5.

7. As the plurality of business support programs, it includes a public program that processes publicly available data and a non-public program that processes non-public data that cannot be made public, The task processing unit, When the public program is selected by the selection unit, based on the public program and the public data, it processes the task corresponding to the input data, When the non-public program is selected by the selection unit, based on the non-public program and the non-public data, it processes the task corresponding to the input data. The business support system according to claim 1 or 2.

8. The selection unit selects the AI that can select the business support program. The business support system according to claim 1 or 2.

9. When a plurality of the business support programs are selected by the selection unit, the task processing unit processes the task corresponding to the input data based on the plurality of business support programs and merges the outputs from each of the plurality of business support programs. The business support system according to claim 1 or 2.

10. When a plurality of the business support programs are selected by the selection unit, the task processing unit processes the task corresponding to the input data by inputting the output from some of the plurality of business support programs into other of the plurality of business support programs. The business support system according to claim 1 or 2.

11. Obtain input data indicating a user's input, Based on an AI (Artificial Intelligence) that can select the business support program specialized for the task corresponding to the input data among a plurality of business support programs specialized for specific tasks in business support, the business support program can be selected, When the business support program is selected by the selection unit, based on the business support program, it processes the task corresponding to the input data. Business support method.

12. An input data acquisition unit that acquires input data indicating a user's input, Among a plurality of business support programs specialized for specific tasks in business support, a selection unit capable of selecting the business support program specialized for the task according to the input data based on AI (Artificial Intelligence) that can select the business support program, When the business support program is selected by the selection unit, a task processing unit that processes the task according to the input data based on the business support program, A program for causing a computer to function as such.

Images