Prediction information generation system

Abstract

To provide a prediction information generation system capable of automatically generating prediction information for predicting the future in which created ideas can be recognized.SOLUTION: A prediction information generation system includes: prompt input means for inputting a prompt related to an idea into a foundation model; and story information acquisition means for acquiring story information related to a business story using the idea which is output from the foundation model to the prompt input by the prompt input means.SELECTED DRAWING: Figure 6

Description

Technical Field

[0001] The present invention relates to a prediction information generation system that uses a foundation model to automatically generate prediction information for predicting a future in which ideas created can be appealed.

Background Art

[0002] Patent Document 1 proposes a problem-solving support system that can dig deep into many surrounding deformation forms from the seeds of the created invention and develop them, and further strengthen the technology group by reflecting information on market needs.

[0003] Patent Document 2 proposes a creation support program suitable for supporting the creation of innovation by effectively presenting specific examples in which a solution concept is reflected to the creator.

[0004] In the disclosed technologies of these Patent Documents 1 and 2, when learning technical information such as patent information by artificial intelligence and inputting information on problems to be solved and problems related to an invention to be newly created, an optimal solution concept is proposed via artificial intelligence.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0006] However, there has been no previously proposed creative support tool that focuses on the generated solution concepts or the commercialization of inventions. If commercialization is in mind, if it were possible to obtain prediction information that predicts the future in which the idea of the invention can be appealed, it would be possible to consider in advance how to develop the idea in the future and connect it to business, but the method itself has not been proposed until now.

[0007] Therefore, the present invention has been devised in view of the above-described problems, and its object is to provide a prediction information generation system capable of automatically generating prediction information that predicts the future in which the created idea can be appealed by using a base model (a large-scale model trained with a large amount of data and adaptable to various tasks).

Means for Solving the Problems

[0008] The prediction information generation system according to the present invention includes a prompt input means for inputting a prompt regarding an idea into a base model in which future prediction information predicted about the future is learned, and information acquisition means for acquiring prediction information predicting the future in which the above-mentioned idea can be appealed, which is output from the above-mentioned base model with respect to the prompt input to the above-mentioned prompt input means.

Effects of the Invention

[0009] According to the present invention, it becomes possible to automatically generate prediction information that predicts the future in which an idea can be appealed by using a base model (a large-scale model trained with a large amount of data and adaptable to various tasks).

Brief Explanation of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

[0011] Hereinafter, an idea evaluation system to which the present invention is applied will be described in detail with reference to the drawings.

[0012] First Embodiment FIG. 1 is a configuration diagram schematically showing the configuration of a network including an idea evaluation system 1 to which the present invention is applied. As shown in the figure, the server 10 is communicably connected to the user terminal 30 via a communication network 20 such as the Internet. In FIG. 1, only one user terminal 30 is shown, but the server 10 may be communicably connected to two or more user terminals 30. The server 10 provides an evaluation result designated by the user to the user who operates the user terminal 30.

[0013]

[0014] The computer processor 11 is configured as a CPU, GPU, or the like, reads various programs stored in the storage 15 or the like into the memory 12, and executes various instructions included in the programs. The memory 12 is constituted by, for example, DRAM or the like.

[0015] The input / output I / F 13 includes various input / output devices for exchanging information with an operator or the like. The input / output I / F 13 includes, for example, information input devices such as pointing devices such as a keyboard, a mouse, and a touch panel, voice input devices such as a microphone, and image input devices such as a camera. The input / output I / F 13 also includes image output devices such as a display and voice output devices such as a speaker.

[0016] The communication I / F 14 is implemented as hardware such as a network adapter, various communication softwares, or a combination thereof, and is configured to enable wired or wireless communication via a communication network 20 or the like.

[0017] The storage 15 is constituted by, for example, a magnetic disk or a flash memory. The storage 15 stores various programs including an operating system and various data. For example, as shown in FIG. 1, the storage 15 has an evaluation information management table 152 or the like for managing evaluation information to which an evaluation related to an idea is attached.

[0018] The storage 15 also stores a server-side program 40 according to an embodiment of the present invention. The program 40 is a program for causing the server 10 to function as all or part of a system for providing a service of an idea evaluation system. At least a part of the server-side program 40 may be configured to be executed on the user terminal 30 side via a terminal-side program 42 described later.

[0019] In this embodiment, the server 10 may be configured using a plurality of computers each having the above-described hardware configuration. In such a case, the server 10 is composed of a plurality of server devices.

[0020] The server 10 configured as described above can be configured to have functions as a web server and an application server, execute various processes in response to requests from the user terminal 30, and transmit screen data and control data composed of HTML data or the like according to the results of the processes to the user terminal 30. On the user terminal 30, a screen based on the received data is output.

[0021] Next, the hardware configuration of the user terminal 30 will be described. The user terminal 30 is configured as a general computer and includes, as shown in FIG. 1, a computer processor 31, a memory 32, an input / output I / F 33, a communication I / F 34, and a storage 35. These components are electrically connected via a bus or the like (not shown).

[0022] The computer processor 31 is configured as a CPU, GPU, or the like, reads various programs stored in the storage 35 or the like into the memory 32, and executes various instructions included in the programs. The memory 32 is composed of, for example, DRAM or the like.

[0023] The input / output I / F 33 includes various input / output devices for exchanging information with an operator or the like. The input / output I / F 33 includes, for example, information input devices such as a keyboard, a mouse, a pointing device such as a touch panel, a voice input device such as a microphone, and an image input device such as a camera. The input / output I / F 33 also includes an image output device such as a display and a voice output device such as a speaker.

[0024] The communication I / F 34 is implemented as hardware such as a network adapter, various communication software, and combinations thereof, and is configured to enable wired or wireless communication via a communication network 20 or the like.

[0025] The storage 35 is constituted by, for example, a magnetic disk or a flash memory. The storage 35 stores various programs including an operating system and various data. Programs stored in the storage 35 can be downloaded and installed from an application market or the like. Further, the storage 35 stores the above-described terminal-side program 42. The program 42 is configured as a web browser or other application, and as described above, can be configured to execute at least a part of the server-side program 40.

[0026] Note that the user terminal 30 is configured as a mobile phone, a smartphone, a tablet terminal, a wearable terminal, a personal computer, or the like.

[0027] A user who operates the user terminal 30 configured as described above can utilize the service of the idea evaluation system provided by the server 10 by executing communication with the server 10 via the terminal-side program 42 installed in the storage 35 or the like.

[0028] Next, the functions of the server 10 configured as described above will be described. As shown in FIG. 1, the computer processor 11 of the server 10 is configured to function by executing a program read into the memory 12 or instructions included in at least a part of the server-side program 40.

[0029] Then, server 10 acquires evaluation information based on the information received from user terminal 30 or the information generated on the server 10 side. At this time, server 10 inputs a prompt regarding the created idea into the base model, and acquires evaluation information to which an evaluation regarding that idea generated by the base model is attached. The base model is an AI neural network trained with a large amount of unlabeled dataset, and is a type of machine learning model trained to execute various tasks from text translation to medical image analysis. The base model is programmed to function by understanding general contexts of patterns, structures, and expressions. Based on this basic understanding of communication methods and pattern identification methods, a knowledge baseline is created. By further modifying or fine-tuning this, it is possible to execute tasks in specific areas of almost all industries. Many base models, especially those used in natural language processing (NLP), computer vision, and audio processing, are pre-trained using deep learning techniques. The base model may be a model trained to be able to generate, identify, classify, predict, translate, etc. natural language, still images, moving images, and music by performing transfer learning, fine-tuning, and distillation. The base model may be a model trained to be able to detect information that is ethically problematic, discriminatory and harmful, factually baseless, or simply useless. The base model may be a model trained to be able to detect factually baseless information in natural science using a natural science simulator. Since the base model may include hallucination, it may be a model fine-tuned to detect hallucination by reinforcement learning from human feedback (RLHF: Reinforcement Learning from Human Feedback). The base model may be a model that reflects "human value criteria" such as human preferences and intentions by reinforcement learning from human feedback.

[0030] Base models trained with natural language processing can recognize context, grammar, and language structure, and generate and extract information from the data used for training. A base model trained with natural language processing is, for example, a learned model that performs natural language processing according to an instruction text and generates a response text. The instruction text may at least include information about the created idea. The base model may be an interactive, so-called chat-type or conversation-type model that alternately receives the instruction text and generates the response text. The natural language model may be a large language model (LLM) trained with a large amount of text data, or a model obtained by transfer learning of the large language model. Also, the natural language model is not limited to being constructed within the server 10, and may be constructed in a system external to the idea evaluation system 1. Further, the base model may be a diffusion model used for image generation AI.

[0031] In addition, M&A information regarding partnerships, mergers, and acquisitions between companies may be learned in this base model. The M&A information referred to here may be information on which companies have collaborated or partnered in business, for example, from news articles on the Internet, or information regarding mergers and acquisitions between companies. Also, by analyzing patent information, for companies that have filed joint applications, they may be regarded as being in a partnership relationship and included in the M&A information.

[0032] In addition, the base model may be trained with future prediction information predicted about the future. The future prediction information may be, for example, information on publicly known future predictions published on the Internet under titles such as "Predicting the future ● years later", "Megatrends in the world ● years later", and "Scenarios of society / companies ● years later". Since these future prediction information are predicted by experts based on data, etc. by government agencies, think tanks, various industry groups, and private research companies, they may have high credibility, and this can be incorporated and learned by the base model. In addition to information on the Internet, the future prediction information may be incorporated by reading media such as books and materials and converting them into electronic data. Also, if these books are sold or provided as electronic data, they may be incorporated as they are. These M&A information and future prediction information may be incorporated and learned, for example, based on RAG (Retrieval-Augmented Generation).

[0033] A large language model is a deep learning model that pre-learns from a vast amount of data a language model called a model that models human spoken language by its probability of occurrence. That is, a large language model is a natural language processing model trained using a large amount of text data, which takes an article as input and outputs an article. When a large language model is applied to a system that conducts question-and-answer sessions, when a question sentence is input into the large language model, an answer sentence is output from the large language model.

[0034] When server 10 receives the text data (prompt) of a request, it uses a large language model to statistically estimate the generation probability of the next word from the text included in the received prompt, and sends the estimation result to the requester. As the large language model, for example, as an Internet site, known technologies described in "https: / / huggingface.co / datasets / databricks / databricks-dolly-15k", "https: / / huggingface.co / datasets / togethercomputer / RedPajama-Data-1T", etc. can be adopted. Also, as the large language model, for example, GPT-4 provided by OpenAI in the United States may be used.

[0035] As shown in FIG. 2, a prompt regarding an idea is input. This prompt regarding an idea is composed of text data describing the content of the idea. This prompt may be composed of each item such as the summary of the invention (for example, "adopting a vertical axis wind turbine to capture wind energy from all directions", etc.), the problem to be solved (for example, "in the conventional horizontal axis wind turbine, the wind hitting is limited", etc.), the solution method (for example, "introducing a vertical axis wind turbine that captures wind energy from all directions", etc.), the effect of the invention (for example, "improvement in flexibility with respect to changes in the wind direction", etc.), or may be, for example, something that expresses the invention concept or the solution method in one word such as "make it a nested structure".

[0036] This prompt may be something thought up by the user through their own creative activities, or it may be based on ideas generated by artificial intelligence based on well-known techniques shown in, for example, Patent Documents 1 and 2. In such a case, by pre-learning patent information, a basic model for idea creation that searches for solutions for the input text data is prepared in advance. This basic model for idea creation may utilize an LLM. Then, by inputting text data such as "How can the power generation effect of wind power generation be enhanced?" into the learned model, ideas consisting of the outline of the invention, problems to be solved, solutions, effects of the invention, etc. are output. The ideas output may express the invention concept or solution method in one sentence. In outputting such ideas, it may be output in text data, but it is not limited to this, and the content of the output text data may be illustrated or the drawn content may be displayed. In such a case, a well-known generative AI may be used, and a basic model that illustrates and outputs the content to be output as an alternative to text for the input text data may be utilized.

[0037] The text information of ideas created by the user themselves or output from the learned model is input as a prompt. The input of the prompt may be via the user terminal 30, or if it has already been stored in the storage 35 of the user terminal 30, that may be utilized. Alternatively, it may be input on the server 10 side, or if it has already been stored in the storage 15 of the server 10, that may be utilized.

[0038] Particularly for ideas generated by the learned model, since they may already be stored in the user terminal 30 or the server 10, they may be utilized as they are.

[0039] Next, the input prompt is input to the large language model to obtain evaluation information with an evaluation regarding the idea. This operation is basically performed via the large language model on the server 10 side. Therefore, the prompt obtained at the user terminal 30 is to be transmitted to the server 10 via the communication network 20. The server 10 may display the evaluation information obtained via the large language model via the input / output I / F 13, or may transmit the evaluation information to the user terminal 30 via the communication network 20 and display it via the input / output I / F 33 at the user terminal 30.

[0040] In addition to being embodied as such an idea evaluation system 1, the present invention may also be embodied as a program capable of implementing these functions.

[0041] In addition, when the function of generating evaluation information based on the large language model for the input prompt is provided on the server 10 side, these operations are executed via the storage 15 and the server-side program 40. In addition, the function of generating evaluation information based on the large language model for the input prompt may be provided not on the server 10 but on the user terminal 30 side. In such a case, these operations are executed via the storage 35 and the terminal-side program 42 in the user terminal 30. Then, the prompt obtained at the user terminal 30 is input directly into the large language model within this user terminal 30 without being transmitted to the server 10.

[0042] In addition, as the evaluation regarding the idea, any one or more of the resources or costs required to provide a product or service including the idea to customers, the marketability of the idea, and the customer satisfaction of the idea may be included.

[0043] Details, outlines, specifications, manufacturing methods, manufacturing man-hours, required materials, import and export destinations, information on logistics necessary for transporting products and materials, etc. of products or services including ideas can also be obtained through large language models. After searching for this information once, the resources and costs required to provide the product or service to customers may be calculated from unit prices prepared in advance as templates from manufacturing man-hours, labor costs, material costs, etc., or these resources and costs may also be obtained through large language models, or the resources and costs required to directly provide the product or service to customers may be obtained from the input prompt. Also, the evaluation of costs and resources may be performed by preparing existing cost calculation programs, man-hour calculation programs, etc. in advance and using the results calculated by inputting parameters into these programs.

[0044] And in addition to simply outputting the necessary resources or costs through this large language model, from the quantity of the resources and the magnitude of the costs, the actual business potential itself may be evaluated and included in the evaluation information.

[0045] Marketability indicates information regarding the possibility that the idea will be accepted by customers in the market, in other words, the possibility that customers will purchase the idea, the share that can be obtained in the market, etc. The marketability of an idea is determined and the result is output through a large language model from a prompt regarding the idea, and based on information on the Internet as necessary.

[0046] At this time, a market simulator 51 that performs a marketability simulation as shown in FIG. 3 may be used.

[0047] The market simulator 51 is a tool that performs the above-described simulation regarding marketability based on the input parameters. This market simulator 51 may utilize any existing tool. Examples of the information implemented in the market simulator 51 include financial statements of listed companies, expert comments and opinions regarding futures predictions and reasons for increases or decreases in stocks published in analyst reports, newspaper articles, etc. Another example of the information implemented in the market simulator 51 may include expert opinions regarding the entire Nikkei average futures, or opinions regarding specific segments, industries, and furthermore opinions regarding individual futures, taking in comments and upward or downward predictions by experts (analysts) published on the Internet. Also, examples of the information implemented in the market simulator 51 may include information such as movements in interest rates, futures, foreign exchange rates, stock prices of each brand, crude oil, precious metals, Bitcoin, etc., serial charts, line graphs, Bollinger Bands, trading volume, MACD, moving averages, etc. Examples of the information implemented in the market simulator 51 may include those containing various data regarding politics, economy, society, technology, etc., such as GDP, employment statistics, mining and industrial production index, capital investment, labor force survey, business sentiment index, consumer spending, number of new car sales, consumer price index, etc.

[0048] The market simulator 51 performs a marketability simulation based on a well-known simulation method based on parameters according to the input prompt or information based on the ideas output from the large language model. And it may be possible to obtain evaluation information in which the marketability of the idea is evaluated based on the result of the simulation by this market simulator 51. At this time, regarding how much it will be accepted in the market, the simulation may be performed while referring to examples of markets, similar products, and services similar to the idea. In addition, the market simulator 51 may also refer to data on future predictions of the future market, society, politics, economy, and technology, and simulate marketability. In addition, the market simulator 51 may determine the rarity, etc. based on how many products and services similar to the idea there are, and simulate the marketability from the determination result.

[0049] In addition to outputting the simulation result itself output from the market simulator 51 as evaluation information, the large language model may perform a new solution search based on the simulation result and output evaluation information. In such a case, evaluation information is obtained by inputting a prompt reflecting the result of the simulation by the market simulator 51 into the large language model.

[0050] Regarding the customer satisfaction of the idea, it may be evaluated while referring to examples of markets, similar products, and services similar to the idea. At this time, it may be possible to determine from examples of markets, similar products, and services similar to the idea how much feedback there has been from customers in the past, or from data on sales and satisfaction. In such a case, based on the information posted on SNS and Internet information, the degree of empathy may also be referred to, for example, the number of posts with the intention of empathy such as "like" on the Internet.

[0051] Regarding the evaluation of ideas, it may include the feasibility of the ideas. The feasibility here refers to whether it is technically feasible, but it is not limited to this. It may also include whether it is viable in terms of business models, or the possibility of being infeasible due to factors other than technology such as politics, economy, and society. In such cases, the evaluation of the feasibility of the ideas is carried out by a large language model that has learned information about politics, economy, society, etc.

[0052] Among them, when evaluating the feasibility based on whether it is technically feasible or not, the evaluation of the feasibility of the ideas is carried out by a large language model that has learned information about natural science. The information about natural science here includes all papers in the natural science field, non-patent documents including those published in academic conferences and preprints of international conferences, patent documents, and furthermore, all technical information publicly available on the Internet.

[0053] At this time, as shown in FIG. 3, a natural science simulator 52 that performs natural science simulations may be used.

[0054] The natural science simulator 52 is a tool that performs the above-described natural science simulations based on the input parameters. This natural science simulator 52 may use any existing tool. As the natural science simulator 52, for example, an FEM simulator that analyzes stress fields and strains based on the finite element method, chemical reaction simulations, simulations of chemical engineering including plants, electrical circuit simulations, fluid dynamics simulations, quantum mechanics simulations, etc., tools that can simulate physical phenomena, chemical phenomena, and thus all natural sciences are used.

[0055] The natural science simulator 52 performs simulations of natural science based on well-known simulation methods according to parameters corresponding to the input prompt or the ideas output from the large language model. And through this natural science simulator 52, it may be determined whether the idea itself can be technically realized. For example, if it violates the law of universal gravitation that violates the laws of nature, that fact may be added to the evaluation.

[0056] In addition to whether it is naturally scientifically feasible, if it is naturally scientifically feasible, the magnitude and feasibility of its effects may also be determined via the natural science simulator 52.

[0057] In addition to outputting the simulation result itself output from the natural science simulator 52 as evaluation information, the large language model may perform a new solution search based on the simulation result and output evaluation information. In such a case, evaluation information is obtained by inputting a prompt reflecting the result of the simulation by the natural science simulator 52 into the large language model.

[0058] According to the present invention, evaluation information with an evaluation regarding the novelty or inventiveness of the above idea may be obtained by a large language model that has learned patent information. That is, patent information includes not only all text data of published published gazettes and patent gazettes (hereinafter, these are collectively referred to as patent gazettes), but also patent classifications. In the large language model, related patent gazettes are searched based on the obtained prompt. Then, the patent information described in the searched known patent gazettes is compared with the content of the prompt of the obtained idea. And it is determined whether there are differences between the patent information described in the known patent gazettes and the content of the prompt of the idea, and the degree of the differences, and an evaluation of novelty or inventiveness is performed based on the determination result.

[0059] In such a case, by pre-training the patent information, a pre-established learned model for searching for similar patent gazettes for the input text data may be provided. Then, the input prompt is input into the learned model. As a result, similar patent gazettes (patent documents) for the input prompt are searched from the learned model. That is, text data of similar patent gazettes for the created idea can be obtained from the learned model. Next, it is determined whether there is a difference between the patent information described in the known patent gazette and the content of the idea prompt, and the degree of the difference, and an evaluation of novelty or inventiveness is performed based on the determination result. The searched patent gazette (patent document) may be obtained as one piece of evaluation information.

[0060] At this time, the prompt may be input into a large language model, and text data based on the idea output therefrom may be obtained once. The text data based on the idea is information related to the idea organized through the large language model from the prompt, or is generated in a language suitable for solution search. An evaluation of novelty or inventiveness may be performed based on the text data of the known patent gazette output by inputting this into the learned model.

[0061] In the present invention, as shown in FIG. 4, request information for ideas to be newly created may be generated based on the acquired evaluation information. This request information serves as a prompt (input text shown in FIG. 4) for generating ideas by artificial intelligence, which utilizes a learned model to search for solutions to the input text data. If better request information can be generated based on the evaluation information in this way, the artificial intelligence (learned model) can generate better ideas accordingly. In particular, when the evaluation result is not satisfactory, request information necessary for further improving the invention can be efficiently generated. For example, in the case of an evaluation result indicating that the idea's solution method, "multiple blades and a rotor," lacks novelty and inventiveness, as request information necessary for further improving "responding to wind speed," the following can be generated: "In a wind power generation that combines a low-speed cross-flow type windmill and a high-speed Darrieus type windmill according to wind speed, what is the method for enhancing the power generation effect of wind power generation?" For example, in cases where there is a lack of novelty and inventiveness, the idea can be further improved by generating request information necessary for achieving inventiveness.

[0062] Note that this request information may be input into a basic model for idea creation that uses a large language model. And information regarding the ideas to be newly created, which is output from the large language model for the input request information, may be output. That is, a large language model may be applied when generating ideas by artificial intelligence.

[0063] Also, a large language model may be used when generating request information from the evaluation information. That is, as shown in FIG. 5, the generated evaluation information is input into the large language model. And request information is output from the large language model for the input evaluation information. That is, a large language model may also be applied when generating request information from the evaluation information.

[0064] Thus, according to the present invention, by utilizing a large language model, the created ideas can be automatically evaluated from various perspectives such as feasibility, marketability, patentability, etc. As a result, it is possible to automatically evaluate the solution concepts and inventions completed with the help of artificial intelligence, organize them when a large number of solution concepts and inventions are automatically generated by artificial intelligence, and easily realize the work of summarizing them into the necessary technologies for linking to the value provided to customers.

[0065] Also, even when the evaluation result is not satisfactory, the large language model can generate the request information necessary to further improve the invention.

[0066] Note that the present invention is not limited to the above-described embodiments. For example, based on the evaluation information, companies that promote the utilization of the idea may be searched for.

[0067] The companies that can utilize the idea refer to the companies that actually want to use the idea and the companies that can gain benefits through partnerships through open innovation. In searching for such companies, the search is conducted based on public information, that is, the content of the patent information and non-patent documents of each company, the information on the company's website and news releases. The search itself may be carried out using artificial intelligence, or more preferably, using a large language model. In such a case, the evaluation information is input into the large language model, and the large language model outputs the companies that promote the utilization of the idea based on the above-described various information.

[0068] Note that based on the above-described various evaluation information, the acquired story information may be modified. In such a case, using the well-known LLM technology, the amendment to the story information may be listed as a correction candidate, or the story information itself may be output from the beginning as the one modified based on this evaluation information.

[0069] Regarding the correction of story information based on evaluation information, for example, story information with higher evaluation content in the evaluation information may be generated based on the LLM and proposed.

[0070] Furthermore, as evaluation information, evaluation information obtained by evaluating a story from a moral perspective by a base model that has learned information on moral knowledge may be used. Information on moral knowledge consists of, for example, data that describes information that serves as a model in terms of actions, thoughts, and philosophy from a moral perspective and an ethical perspective. Examples of information on moral knowledge may include those who describe ethical dilemmas in text form and provide interpretations and countermeasures from different perspectives. For example, in response to the question "Scenario: The train is out of control and there are 5 people standing in the direction of travel. By switching the switch, the train can be made to go on another track, but there is 1 person standing on that track. Do you switch the switch?", information such as "A: Switch the switch (utilitarian perspective that emphasizes results)", "B: Do not switch the switch (Kantian perspective that emphasizes the morality of actions)" may be learned. Also, for example, in response to the question "What criteria should an autonomous vehicle use to avoid an accident?", data that has learned the answer "Ethical perspective: Should the safety of pedestrians be prioritized or should the passengers be protected?" may be used. In addition to answers to such questions, learning data that has learned documents that briefly and specifically explain each moral theory, ethics, and philosophical approach may be accumulated, and a base model that has learned information on moral knowledge may be prepared based on this. Data that describes ethical dilemmas in text form and shows interpretations and countermeasures from different perspectives may also be learned.

[0071] Second Embodiment Hereinafter, the second embodiment will be described. In the second embodiment, the configuration of the idea evaluation system 1 described in the first embodiment above is used as it is, and the processing operations described in the first embodiment are appropriately combined and executed.

[0072] In the second embodiment, a prompt related to an idea is input into the base model, and story information related to a business story that makes use of the idea output from the base model for the input prompt is obtained.

[0073] A business story that makes use of an idea indicates a context consisting of a stage or characters, a conflict, and a big idea that leads to a solution constituted by the idea, as shown in FIG. 6.

[0074] The stage or characters indicate the situations and backgrounds necessary for leading to the idea, that is, the solution. The characters are the actual characters appearing in the story, and the characters are those with personalities that the listeners such as customers, business partners, partner companies, employees, and stakeholders can empathize with.

[0075] The context is formed through this stage or characters. For example, in the case of an idea (solution) related to two-dimensional code technology, a story (text information) consisting of a stage of the two-dimensional code payment market is generated, and the characters include the party proposing this solution (e.g., a company developing and producing two-dimensional codes). Note that the characters may be omitted in forming this context.

[0076] And the conflict indicates what problems or unfortunate events may occur if the current situation remains unchanged. For example, in the case of an idea (solution) related to two-dimensional code technology, a story is generated that causes a conflict in the mind such as that if the current two-dimensional code continues to be used, various incidents will occur due to security problems, and something must be done to get out of this situation. That is, through the conflict, excitement is created, a strong impression that cannot be forgotten by the other party is implanted, getting away from a boring story, and the other party will be interested and listen intently.

[0077] A big idea is not the solution itself, but rather a higher-level conceptualization of the solution or a summary of the solution expressed in a catchphrase or the like. Create a catchphrase that will soothe the listener and leave a lasting impression when they hear the big idea. This big idea may be generated, for example, based on the context and conflicts generated in the previous stage in addition to the idea. Examples of big ideas include "Introducing a system for measuring biological data 〇〇 without needles," "Building an 〇〇 investment platform that is easy for 〇〇 investors to use," and "Providing a 〇〇 prevention function in the two-dimensional code itself," which may convey what one wants to do in one sentence.

[0078] In generating such a story, it is also possible to propose through the generated story how the generated idea should be linked to business and monetized. Also, a suitable business model for realizing such a story may be generated and proposed.

[0079] An example of the generated story will be described. For example, if the prompt is "Miniaturize the structure of a fuel cell," a story for commercialization could be "Development and sales of ultra-small fuel cells for wearable devices," and the details could be "Utilize nanoscale porous separator technology to develop an ultra-small fuel cell that realizes a driving time more than 10 times that of conventional lithium-ion batteries. By providing it for wearable devices such as smartwatches and headsets, create a market for next-generation wearable devices that can be used for a long time. Furthermore, develop a regular delivery service for fuel cartridges to build a continuous revenue model." and so on.

[0080] If M&A information is learned in the base model, story information that takes these into account will be generated. Therefore, especially for users considering M&A such as partnerships, mergers, and acquisitions between companies, they can obtain a story that takes into account matters related to that M&A while making use of the idea.

[0081] Similarly, if future prediction information has been learned, story information that takes these into account will be generated. While leveraging ideas, it is possible to obtain a story that takes into account predicted future developments in politics, economy, society, technology, etc.

[0082] Also, the feasibility may be evaluated through the natural science simulator described above, and the evaluated information thus informatized may be attached to the generated and proposed story.

[0083] When proposing this story, the evaluation information shown in the first embodiment described above may also be shown.

[0084] Also, simulation may be performed using the market simulator in the first embodiment, and story information may be obtained based on the results of the simulation. In such a case, if the product or service is not likely to sell well as a result of the simulation by the simulator, an attempt may be made to generate a story in which a more powerful impactful collision occurs.

[0085] Also, by inputting a prompt reflecting the results of the simulation by the market simulator into the base model, a new story may be generated according to the results of the simulation. At this time, if an evaluation regarding the novelty or inventiveness of the idea is obtained, story information based on this may be obtained. For example, if there is no inventiveness, a context corresponding thereto may be created.

[0086] Also, based on the obtained story information, request information for ideas to be newly created may be generated. That is, if there is a further solution in the obtained story and it leads to a more effective solution to the conflict, request information corresponding thereto may be generated.

[0087] In this second embodiment, as shown in FIG. 6, as a basis, information regarding the evaluation of the novelty or inventiveness of an idea, or patent information output in the first embodiment may be shown. Thereby, it is possible to show that the solution, that is, the idea itself has patentability, and it is also possible to imply that the entire generated business story is protected by a patent.

[0088] Note that the present invention is not limited to the above-described embodiments. As a prompt to be input to the base model, instead of a prompt regarding an idea, a prompt regarding a technical need may be input.

[0089] The prompt regarding a technical need here may be, for example, one that is limited in detail from a wide range such as NFT technology, AI technology, nanotechnology, and metaverse technology, to specific nanotechnology such as nanoimprint technology and nanorobot technology.

[0090] Then, need information regarding business needs that make use of the above technical needs output from the base model for a prompt regarding such technical needs is acquired. This need information is information indicating in which actual marketing segment and customer layer there is a need for a specific offering value that makes use of the technical need.

[0091] Conversely, when a prompt regarding need information is input, information regarding technical needs may be output from the base model.

[0092] Note that in both the first embodiment and the second embodiment, the prompt to be input to the base model is not limited to text data, and image data may be input. In such a case, text information may be extracted from the image data using well-known techniques, or data obtained by vectorizing the semantic content of the image data may be input to the base model as a prompt.

[0093] Similarly, in both the first and second embodiments, the information output from the base model is not limited to text data, sentences, or character strings, and may output diagrams, tables, image data, etc. Especially when generating story information, not only text data but also output as an image may make the story easier to understand, or a combination of an image and text data is often easier to understand.

[0094] Third Embodiment Hereinafter, the third embodiment will be described. In the third embodiment, the configuration of the idea evaluation system 1 described in the above-described first and second embodiments is used as it is, and the processing operations described in the first embodiment are appropriately combined and executed.

[0095] In the third embodiment, a prompt regarding an idea is input to the base model, and prediction information is obtained that predicts the future form in which the above-mentioned idea output from the base model for the input prompt can be appealed.

[0096] In this third embodiment, it is desirable that future prediction information predicted about the future is learned in the base model.

[0097] In the third embodiment, instead of outputting an idea from the base model for the input prompt, prediction information is generated that predicts the future society, lifestyle, business, etc. in which the idea can be appealed.

[0098] For example, for an idea such as "Develop a separator with a nanoscale porous structure, optimize the gas flow, and suppress the thickness to less than 1 / 10 of the conventional one. Integrate the separators on the anode side and the cathode side to achieve further miniaturization.", the prediction information is a future society that makes use of a low-power ultra-small fuel cell, such as "A society in which medical devices are constantly implanted in the body for a long time, and health checks can be automatically performed at all times through the medical devices."

[0099] If future prediction information is learned in the base model, it is possible to obtain prediction information that takes into account future predictions while making use of ideas.

[0100] This future prediction information may include information regarding predicted needs for the future. The information regarding such needs includes information regarding all needs such as customer needs and social needs. In such a case, prediction information that predicts needs for which ideas can be appealed is output from the base model.

[0101] As an example of a need for which an idea can be appealed, when the idea to be generated for the prompt "miniaturize a fuel cell" is an idea such as "develop a separator with a nanoscale porous structure, optimize the gas flow, and suppress the thickness to 1 / 10 or less of the conventional one. Integrate the separators on the anode side and the cathode side to achieve further miniaturization.", the generated need (prediction information) is a need such as "need for a lightweight VR device that can be used for a long time".

[0102] At this time, request information regarding matters expected for the object of the idea may be input as a prompt. The request information here is, for example, all matters expected for objects such as "wind turbine", "fuel cell", "concrete structure", "thermal power generation element", etc. for which the idea is to be polished. The request information may be all information expected for the object, such as problems, effects, expectations, and desired functions of the object (method), input as a prompt. Such request information is not essential until it is input as a prompt, and it may simply generate candidates for request information and display them to the user.

[0103] Candidate request information may be generated based on prediction information. In such a case, for example, when text data such as "a society in which a medical device is constantly implanted in the body for a long time and health checks can be constantly and automatically performed through the medical device" is generated as prediction information, phrases such as "long-term implantation in the body" and "automatic health check" may be listed as candidate request information from this. For the generation of such phrases, an LLM may be used as appropriate. As a result, the user can input the request information as a new prompt based on the predicted content suggested by the prediction information, and obtain information on the newly generated ideas output from the base model for this, thereby enabling idea creation.

[0104] In addition, the generated request information may be input to the above-described idea creation base model. In such a case, one or more candidate request information may be generated, and the request information selected by the user may be input to the idea creation base model, or the generated request information may be directly input to the idea creation base model without going through user selection or decision-making to generate ideas. That is, new ideas will be sequentially created by the idea creation base model from the generated request information without involving the user.

[0105] Also, in the third embodiment, corporate information may be input to the base model as a prompt. Examples of this corporate information include company name, year of establishment, business content, location, representative, number of employees, capital, major customers, sales, corporate philosophy / vision, major products and services, history, IR information, etc. Such corporate information is information on the Internet, information read from other recording media, etc.

[0106] In addition to manually inputting this corporate information as a prompt, it may be imported as data. Also, the corporate information may be not only publicly available information but also in-house documents that are confidentially managed within the company.

[0107] In the third embodiment, in addition to the prompt regarding the idea, corporate information is input as a prompt. That is, based on the information on the future vision such as "planning to apply fuel cells to the XX field for 5 years" extracted from the corporate information, in addition to the prompt regarding the desired idea of "wanting to miniaturize the structure of the fuel cell", new ideas may be generated from the base model, or prediction information capable of appealing that idea may be generated. This prediction information is capable of appealing the idea and is a prediction of the future form of the company in the input corporate information.

[0108] In the third embodiment, the above-described market simulator may be utilized. In such a case, the simulation by the market simulator is performed using the input prompt or the prediction information output from the base model as parameters. And the prediction information may be acquired based on the result of the simulation by the market simulator.

Explanation of Signs

[0109] 1 Idea evaluation system 10 Server 11, 31 Computer processor 12, 32 Memory 15, 35 Storage 20 Communication network 30 User terminal 40 Server-side program 42 Terminal-side program 51 Market simulator 52 Natural science simulator 152 Evaluation information management table

Claims

1. Prompt input means for inputting a prompt related to an idea into a base model in which future prediction information predicted about the future has been learned, and information acquisition means for acquiring prediction information that predicts a future in which the idea output from the base model in response to the prompt input to the prompt input means can be appealed. A prediction information generation system characterized by the above.

2. The prompt input means inputs the prompt into a base model in which the future prediction information including information related to needs predicted about the future has been learned, and the information acquisition means acquires prediction information that predicts the needs for which the idea output from the base model can be appealed. The prediction information generation system according to claim 1, characterized by the above.

3. Based on the prediction information acquired by the information acquisition means, the information acquisition means generates candidates for request information regarding matters expected for the target of the idea. The prediction information generation system according to claim 1 or 2, characterized by the above.

4. Request information input means for inputting request information regarding matters expected for the target of the idea into an idea creation base model that has previously learned patent information, and further comprising idea acquisition means for acquiring an idea output from the idea creation base model in response to the request information input to the request information input means, and based on the prediction information acquired by the information acquisition means, the information acquisition means generates candidates for request information to be input to the request information input means. The prediction information generation system according to claim 1 or 2, characterized by the above.

5. Request information input means for inputting request information regarding matters expected for the target of the idea into an idea creation base model that has previously learned patent information, and further comprising idea acquisition means for acquiring an idea output from the idea creation base model in response to the request information input to the request information input means, and based on the prediction information acquired by the information acquisition means, the information acquisition means generates request information to be input to the request information input means and inputs this into the idea creation base model. The prediction information generation system according to claim 1 or 2, characterized by the above.

6. The prompt input means further inputs corporate information into the base model, The above information acquisition means obtains the idea that can appeal to the above idea output from the above basic model for the prompt input to the above prompt input means and the prediction information predicting the future form of the company in the input company information. The prediction information generation system according to claim 1, characterized by the above.

7. Further comprising a market simulator that performs a simulation of its marketability based on the input parameters, The above market simulator performs a simulation using the prompt input to the above prompt input means or the above prediction information output from the above basic model as the above parameters, The above information acquisition means obtains the above prediction information based on the result of the simulation by the above market simulator. The prediction information generation system according to claim 1, characterized by the above.

Images