Information processing device, content generation method, and content generation program

Abstract

Facilitating the generation of content related to objects that perform a given task. [Solution] The information processing device comprises a prompt generation unit that generates a prompt instructing the generation of content related to an object capable of performing a predetermined task, based on relevant information about the object, and a content generation control unit that causes a generation model to generate content using the prompt. The information processing device also makes it possible to optimize the process of generating the above content.

Description

Technical Field

[0001] The present disclosure relates to an information processing apparatus, a content generation method, and a content generation program.

Background Art

[0002] The utilization of AI agents using AI technologies such as generative AI (Artificial Intelligence) has been promoted in various fields. For example, Patent Document 1 below describes a dialogue system that controls an AI agent that interacts with a speaker.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In a situation where the spread of AI agents has advanced further than it is now, it is considered that in order for the user to easily identify each AI agent and to be able to interact with each AI agent with affection, it is necessary to give each AI agent a personality, in other words, to individualize the AI agent.

[0005] For example, it is possible to individualize an AI agent by creating content related to the AI agent, such as an image character that reflects the functions of the AI agent, and providing the content together with the AI agent.

[0006] However, if a company deploys multiple AI agents of the same type, the content associated with those AI agents will be identical, making it impossible to personalize the AI ​​agents within that company. Furthermore, generating corresponding content for each of the multiple AI agents within the company to personalize each agent is not easy in terms of cost and other factors. This problem is not limited to AI agents, but is a common problem that arises when using any object programmed to perform a predetermined task.

[0007] This disclosure has been made in view of the above-mentioned problems, and one exemplary purpose thereof is to provide a technique for facilitating the generation of content related to an object programmed to perform a given task. [Means for solving the problem]

[0008] An information processing device relating to an exemplary aspect of this disclosure includes a prompt generation means that generates a prompt instructing the generation of content relating to an object programmed to perform a predetermined task, based on relevant information relating to the object; and a content generation control means that causes a generation model that generates content corresponding to an input prompt to generate content using the prompt.

[0009] In an exemplary aspect of the present disclosure, a content generation method is performed in which at least one processor generates a prompt that instructs to generate content relating to an object based on relevant information relating to an object programmed to perform a predetermined task, and a content generation control process that uses the prompt to cause a generation model that generates content corresponding to the input prompt to generate content.

[0010] A content generation program relating to an illustrative aspect of this disclosure causes a computer to function as a prompt generation means that generates prompts instructing the computer to generate content relating to an object programmed to perform a predetermined task, based on relevant information relating to said object, and a content generation control means that causes a generation model that generates content corresponding to the input prompt to generate content using said prompts. [Effects of the Invention]

[0011] One exemplary aspect of this disclosure is that it can provide a technique that facilitates the generation of content related to an object programmed to perform a predetermined task. [Brief explanation of the drawing]

[0012] [Figure 1] This is a block diagram showing the configuration of the information processing device related to this disclosure. [Figure 2] This is a flowchart showing the flow of the content generation method related to this disclosure. [Figure 3] This is a block diagram showing the configuration of other information processing devices related to this disclosure. [Figure 4] This figure shows an example of content generation. [Figure 5] This figure shows other examples of content generation. [Figure 6] This figure shows yet another example of content generation. [Figure 7] This figure shows an example of a user interface (UI) screen for accepting user modifications. [Figure 8] This figure shows an example of generating new content by incorporating the characteristics of other content into already generated content. [Figure 9] Figure 3 is a flowchart showing the processing flow executed by the information processing device. [Figure 10] This is a block diagram showing the configuration of a computer that functions as an information processing device related to this disclosure.

Best Mode for Carrying Out the Invention

[0013] Hereinafter, embodiments of the present invention will be illustrated. However, the present invention is not limited to the following exemplary embodiments, and various modifications are possible within the scope shown in the claims. For example, embodiments obtained by appropriately combining the technologies (part or all of the objects or methods) employed in the following exemplary embodiments may also be included in the scope of the present invention. Further, embodiments obtained by appropriately omitting a part of the technologies employed in the following exemplary embodiments may also be included in the scope of the present invention. Also, the effects mentioned in the following exemplary embodiments are merely examples of the effects expected in those exemplary embodiments and do not define the scope of the present invention. That is, embodiments that do not exhibit the effects mentioned in the following exemplary embodiments may also be included in the scope of the present invention.

[0014] 〔First Exemplary Embodiment〕 A first exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. This exemplary embodiment is a basic form for each of the exemplary embodiments described later. Note that the scope of application of each technology employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technology employed in this exemplary embodiment can be employed in other exemplary embodiments included in the present disclosure as long as there are no particular technical obstacles. Also, each technology shown in the drawings referred to for explaining this exemplary embodiment can be employed in other exemplary embodiments included in the present disclosure as long as there are no particular technical obstacles.

[0015] (Configuration of Information Processing Apparatus 1) The configuration of the information processing apparatus 1 according to this exemplary embodiment will be described with reference to FIG. 1. FIG. 1 is a block diagram showing the configuration of the information processing apparatus 1. As shown in FIG. 1, the information processing apparatus 1 includes a prompt generation unit 101 and a content generation control unit 102.

[0016] The prompt generation unit 101 generates a prompt that instructs to generate content related to an object based on related information about the object programmed to be capable of executing a predetermined task.

[0017] The above "object" may be anything programmed to be capable of executing a predetermined task, and may have a physical entity or may not have a physical entity. For example, the above object may be a computer program itself (which can also be referred to as software), or a device (such as a robot) operated by a computer program, etc.

[0018] Also, the above "predetermined task" may be any task. An object capable of executing a predetermined task can also be referred to as an object having a predetermined function. For example, when the above object is a computer program, the predetermined task executed by that object may be, for example, to output the result of performing predetermined information processing using the input in response to a user's input. Also, for example, when the above object is a robot, the predetermined task executed by that object may be, for example, to execute an operation according to a user's input. Also, the above predetermined task may be something executed by the above object alone, or may be something executed together with other objects or users.

[0019] Furthermore, the information processing device 1 has a function to generate content related to the above object, and the generated content is content that gives the object its own personality. From this perspective, it is preferable that the above object interacts with the user through a predetermined interface. For example, the above object may be software called an agent or software agent that performs tasks on behalf of the user. Among agents, an AI agent that utilizes AI (Artificial Intelligence) is suitable as the above object because it can behave in a manner similar to that of a human.

[0020] For example, the above object could be an AI agent that takes the place of a doctor or other medical professional to interview patients online about their symptoms. In this case, for example, the information processing device 1 could generate content (e.g., a character) tailored to each patient's preferences, and the AI ​​agent could conduct interviews with each patient through that content. This is expected to improve the efficiency of the interviews and increase patient satisfaction.

[0021] Furthermore, the "related information" mentioned above can be any information that has some relation to the object and can be used to generate the content. For example, the related information could include a manual describing the tasks performed by the object, a specification sheet for the object, or a list of the object's capabilities or functions. Alternatively, the related information could include information indicating the user (e.g., a company or individual) who will be using the object. As will be explained in more detail later, by using related information that indicates the tasks performed by the object, it is possible to generate content that reflects those tasks. Additionally, by using information indicating the user as related information, it is possible to generate content tailored to that user. In addition, it is also possible to use information (which could be image data, text data, or both) indicating a person (which may or may not be a real person) who will serve as the model for the content to be generated as related information.

[0022] Furthermore, generating a prompt based on relevant information means that relevant information is used directly or indirectly in generating the prompt. For example, the prompt generation unit 101 may generate a prompt that includes all of the relevant information, a prompt that includes some of the information extracted from the relevant information, or a prompt that includes information obtained by analyzing the relevant information. A prompt is an indication of the content of instructions for the generation model described later, and can also be referred to as an instruction statement, directive statement, etc.

[0023] Furthermore, the "content" mentioned above relates to the object and can be any content that can be used to give the object personality. For example, if the object is a computer program, the content may be image data (which may be a moving image or a still image) of a character that personifies the computer program. In this case, the content may also be audio data such as the character's voice, theme song, or background music played while the computer program is in use. In addition, the content may also be text data that shows the character's lines, catchphrases, speech patterns, dialects, etc. Furthermore, the content may be values ​​of various parameters that characterize the character's appearance, personality, behavior, movement habits, etc. (for example, strength, delicacy, etc.). In this case, a character corresponding to the object is generated based on the generated content.

[0024] Furthermore, for example, if the object is a robot, the content may include data showing the robot's appearance (including coloring, etc.), data showing the design of parts and accessories attached to the robot (including clothing, etc.), and values ​​of various parameters that characterize the robot's behavior, etc. In this way, even when the object is a tangible object such as a robot, it is possible to generate content that gives the object its own personality.

[0025] The content generation control unit 102 uses the prompt generated by the prompt generation unit 101 to cause a generation model that generates content corresponding to the input prompt to generate content.

[0026] The above-mentioned "generative model" can be any model generated by machine learning to generate content corresponding to the input prompt. The applicable generative model should be one that matches the format of the input prompt and the format of the content to be generated. For example, if the prompt generation unit 101 generates a text-formatted prompt, a generative model that can accept text-formatted prompts should be used. If image data is to be generated as the above content, a generative model trained to generate image data (e.g., DALL E2) should be used. If text data is to be generated as the above content, general-purpose language models such as BERT (Bidirectional Encoder Representations from Transformers) or GPT (Generative Pre-trained Transformer) can also be used as the above generative model.

[0027] As described above, the information processing device 1 according to this exemplary embodiment employs a configuration comprising: a prompt generation unit 101 that generates a prompt instructing the generation of content related to an object programmed to perform a predetermined task based on relevant information about the object; and a content generation control unit 102 that uses the prompt generated by the prompt generation unit 101 to cause a generation model that generates content corresponding to the input prompt to generate content.

[0028] With the above configuration, the user can generate content related to an object simply by providing the information processing device 1 with relevant information about the object programmed to perform a predetermined task. Therefore, the above configuration has the effect of making it possible to easily generate content related to an object programmed to perform a predetermined task. Furthermore, the information processing device 1 makes it possible to simplify or optimize the work of generating content related to an object, which was conventionally done by designers and others and required a great deal of time and expense.

[0029] (Content generation program) The functions of the information processing device 1 described above can also be implemented by a program. In this exemplary embodiment, the content generation program causes the computer to function as a prompt generation means that generates prompts instructing the computer to generate content related to an object programmed to perform a predetermined task, based on relevant information about the object. The content generation control means causes a generation model that generates content corresponding to an input prompt to generate content using the prompts generated by the prompt generation means. This content generation program has the effect of facilitating the generation of content related to an object programmed to perform a predetermined task.

[0030] (Content generation process) The flow of the content generation method according to this exemplary embodiment will be explained with reference to Figure 2. Figure 2 is a flowchart showing the flow of the content generation method. Note that the entity executing each step in this content generation method may be a processor provided in the information processing device 1, a processor provided in another device, or the entity executing each step may be a processor provided in a different device.

[0031] In S1 (prompt generation process), at least one processor generates a prompt that instructs it to generate content related to an object programmed to perform a predetermined task, based on relevant information about that object.

[0032] In S2 (Content Generation Control Processing), at least one processor uses the prompt generated in S1 to cause a generation model that generates content corresponding to the input prompt to generate content.

[0033] As described above, the content generation method according to this exemplary embodiment employs a configuration in which at least one processor performs a prompt generation process that generates a prompt instructing the generation of content related to an object programmed to perform a predetermined task, based on relevant information about the object, and a content generation control process that uses the prompt generated in the prompt generation process to cause a generation model that generates content corresponding to the input prompt to generate content. This content generation method has the effect of making it possible to facilitate the generation of content related to an object programmed to perform a predetermined task.

[0034] [Second exemplary embodiment] A second exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same function as those described in the above-described exemplary embodiment are denoted by the same reference numerals, and their descriptions are omitted as appropriate. The scope of application of each technology adopted in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technology adopted in this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems arise. Furthermore, each technology shown in the drawings referenced to describe this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems arise.

[0035] (Configuration of Information Processing Device 1A) The configuration of the information processing device 1A according to this exemplary embodiment will be described with reference to Figure 3. Figure 3 is a block diagram showing the configuration of the information processing device 1A. The information processing device 1A is a device equipped with the function of generating content related to objects. The information processing device 1A may be a local device used by individual users, or it may be a server that provides content generation services to multiple users.

[0036] In this exemplary embodiment, the example described primarily involves an instance where the object is an AI agent and the content is image data of a character personifying the AI ​​agent (in other words, data showing the character's form). However, the object is not limited to an AI agent, and the content is not limited to character images. In the following description, "AI agent" can be replaced with any "object," and similarly, "image data," "character," etc., can be replaced with any "content."

[0037] As shown in the figure, the information processing device 1A includes a control unit 10A that controls all parts of the information processing device 1A, and a storage unit 11A that stores various data used by the information processing device 1A. The information processing device 1A also includes a communication unit 12A for the information processing device 1A to communicate with other devices, an input unit 13A that receives input to the information processing device 1A, and an output unit 14A for the information processing device 1A to output data. The control unit 10A includes a prompt generation unit 101A, a content generation control unit 102A, a data acquisition unit 103A, a pre-processing unit 104A, a presentation control unit 105A, and a reception unit 106A.

[0038] The prompt generation unit 101A, similar to the prompt generation unit 101 in Exemplary Embodiment 1, generates prompts that instruct the generation of content relating to an object programmed to perform a predetermined task, based on relevant information relating to that object. Specifically, the prompt generation unit 101A generates prompts that instruct the generation of image data of the AI ​​agent's character, based on relevant information relating to the AI ​​agent.

[0039] The content generation control unit 102A, similar to the content generation control unit 102 in Exemplary Embodiment 1, uses the prompt generated by the prompt generation unit 101A to cause a generation model that generates content corresponding to the input prompt to generate content. Specifically, the content generation control unit 102A causes the generation model to generate image data of the AI ​​agent character. The generation model used by the content generation control unit 102A will be referred to as "generation model M2" below. Below, we will describe an example in which generation model M is a model (text-to-image model) that accepts input of a text-formatted prompt indicating the image data to be generated and outputs the image data. However, generation model M2 is not limited to a text-to-image model; it can be any model generated by machine learning to generate content corresponding to the input prompt, similar to the generation model used by the content generation control unit 102 described in Exemplary Embodiment 1.

[0040] The data acquisition unit 103A acquires various data necessary for content generation. For example, the data acquisition unit 103A acquires relevant information about the AI ​​agent. The method of data acquisition is not particularly limited. For example, the data acquisition unit 103A may acquire data from other devices (e.g., terminal devices used by the user) or a predetermined database via the communication unit 12A, or it may acquire data that is input via the input unit 13A.

[0041] The preprocessing unit 104A performs predetermined preprocessing on the relevant information acquired by the data acquisition unit 103A. Details of the preprocessing by the preprocessing unit 104A will be described later based on Figures 4 and 6. A language model that has already learned natural language is used for the preprocessing. The language model used by the preprocessing unit 104A will be referred to as "language model M1" below. Furthermore, machine learning natural language means, more specifically, learning the arrangement of its constituent elements (words, etc.) in natural language sentences and the arrangement of sentences in texts. Examples of language models that have learned natural language include BERT and GPT.

[0042] The presentation control unit 105A presents the content generated by the control of the content generation control unit 102A to the user. The manner of presentation is not particularly limited and may depend on the data format of the generated content. For example, if the generated content is image data or text data, the presentation control unit 105A may present the content by displaying it on a display device. Alternatively, if the generated content is audio data, the presentation control unit 105A may present the content by outputting it as audio to an audio output device. The various output devices, such as display devices and audio output devices, used for presenting content may be provided by the information processing device 1A or may be external devices.

[0043] The reception unit 106A accepts various operations related to the generation and modification of content. For example, the reception unit 106A accepts operations to modify content presented by the presentation control unit 105A, and operations to instruct the regeneration of content. Content modification will be described later based on Figure 7, and content regeneration will be described later based on Figure 8.

[0044] As described above, the information processing device 1A, like the information processing device 1 of the exemplary embodiment, employs a configuration that includes a prompt generation unit 101A that generates a prompt instructing the generation of content related to an object programmed to perform a predetermined task, based on relevant information about the object, and a content generation control unit 102A that uses the prompt generated by the prompt generation unit 101A to cause a generation model that generates content corresponding to the input prompt to generate content.Therefore, the effect is obtained that it becomes possible to facilitate the generation of content related to an object programmed to perform a predetermined task.For example, with the information processing device 1A, it is also possible to easily generate image data of a character that personifies an AI agent.

[0045] (Content generation example 1: An example of extracting necessary information as a preprocessing step) Figure 4 shows an example of content generation by the information processing device 1A. In the example in Figure 4, image data of an AI agent's character is generated from a manual describing that AI agent. In other words, in the example in Figure 4, the "object" mentioned above is the AI ​​agent, the "content" mentioned above is the image data, and the "related information" mentioned above is the manual describing the AI ​​agent.

[0046] Note that the language model M1 and generation model M2 shown in Figure 4 may be located outside the information processing device 1A (for example, on a server), or the language model M1 and generation model M2 may be stored in the information processing device 1A.

[0047] Generally, an AI agent's manual describes its functions and characteristics, making it possible to use the manual as relevant information for that AI agent. However, manuals often contain many details that are not representative of the AI ​​agent's characteristics. Therefore, if image data is generated by referencing the entire manual, it is possible that the generated image data may not capture the AI ​​agent's characteristics.

[0048] Therefore, in the example shown in Figure 4, the language model M1 is used to extract information from the instruction manual for generating image data. This process is performed by the preprocessor 104A. Specifically, the preprocessor 104A inputs a prompt P1 to the language model M1 instructing it to extract descriptions from the instruction manual of the AI ​​agent that are considered useful for generating the AI ​​agent's character. The prompt P1 can be generated by embedding the description from the instruction manual into a standard template. If there is a limit to the number of characters that can be input to the language model M1 and the entire text of the instruction manual cannot be input to the prompt P1, the preprocessor 104A can divide the instruction manual into multiple parts and extract descriptions from each part.

[0049] In the example in Figure 4, prompt P1 is input to language model M1, and response A1 is output from language model M1. Response A1 lists descriptions from the instruction manual that are considered useful for generating the AI ​​agent's character (more precisely, image data of that character).

[0050] Next, the prompt generation unit 101A generates a prompt using the information extracted by the preprocessing unit 104A. In the example in Figure 4, prompt P2 is generated using answer A1. Prompt P2 is a prompt that instructs the generation of content (specifically, image data of the AI ​​agent character).

[0051] Prompt P2 displays the information extracted by the preprocessor 104A as the "features" of the AI ​​agent. This makes it possible to generate character image data that reflects the AI ​​agent's features as described in the manual. These features also include information indicating that the AI ​​agent is an accounting management agent, that is, that the tasks performed by this AI agent are accounting management tasks. In this way, by using related information that includes information indicating the task, it becomes possible to generate image data that reflects that task.

[0052] Furthermore, prompt P2 indicates the "company using" the AI ​​agent, and includes a statement instructing that the AI ​​agent be used by that company and that the character design be adapted for that company. This makes it possible to reflect the characteristics of the company using the AI ​​agent in the character's image data. Note that the company using the AI ​​agent can be acquired by the data acquisition unit 103A as related information, separate from the instruction manual.

[0053] In the example shown in Figure 4, the content generation control unit 102A inputs prompt P2 to the generation model M2, which in turn outputs image data of character C1. Character C1 is designed to include a calculator, an item that evokes accounting management, as well as a truck, which evokes the transportation company that uses the AI ​​agent. According to the information processing device 1A, by simply providing the AI ​​agent's manual and the company using the AI ​​agent as related information, it is possible to generate image data of such a character that reflects the tasks performed by the AI ​​agent and the company using the AI ​​agent.

[0054] As described above, the preprocessing unit 104A may extract information to be used for content generation from related information using a language model M1 that has been trained on natural language. The prompt generation unit 101A may then generate a prompt using the information extracted by the preprocessing unit 104A. With this configuration, in addition to the effects performed by the information processing device 1, it becomes possible to generate appropriate content from related information that includes both information useful for content generation and information unnecessary for content generation.

[0055] Furthermore, as described above, the related information may include information indicating tasks that the object can perform. In this case, the prompt generation unit 101A may generate a prompt instructing the generation of content corresponding to the above task. This provides the effect of easily generating content corresponding to tasks that the object can perform, in addition to the effects performed by the information processing device 1.

[0056] Furthermore, as described above, the related information may include information indicating the user of the object. In this case, the prompt generation unit 101A may generate a prompt instructing the generation of content corresponding to the user. This provides the effect of easily generating content corresponding to the user of the object, in addition to the effects performed by the information processing device 1.

[0057] Alternatively, the language model M1 may be instructed to generate prompts to be input to the generation model M2. For example, when using a generation model M2 that generates content described by an explanatory text, the prompt generation unit 101A may instruct the language model M1 to generate an explanatory text describing the content in text format. To give a specific example, the prompt generation unit 101A may generate a prompt that includes related information or a description extracted from the related information by the preprocessing unit 104A, and instructs the language model M1 to generate a character description based on the related information or the description, and input it to the language model M1. As a result, for example, a description such as "a character for accounting management agents, a hero-like character that can perform multiple tasks with high speed" will be output from the language model M1. The content generation control unit 102A can then input the description generated in this way as a prompt to the generation model M2.

[0058] Furthermore, if the generation model M2 is a model that can only accept text written in a specific language (for example, text written in English), the prompt generation unit 101A only needs to generate a prompt written in that language from the relevant information. Alternatively, the preprocessing unit 104A may be used to translate the relevant information, and the prompt generation unit 101A may generate a prompt from the translated relevant information. Conventional translation methods, such as using a language model, can be applied as appropriate for the translation.

[0059] Furthermore, the preprocessing unit 104A may generate feature information that indicates the characteristics of the AI ​​agent from the related information. For example, the preprocessing unit 104A may use a feature information generation model that has been trained to recognize the correspondence between natural language sentences or words and feature information that indicates the characteristics of those sentences or words, to generate feature information (e.g., feature vectors) for each sentence or word included in the related information. In this case, the prompt generation unit 101A should generate prompts that instruct the AI ​​agent to generate characters with similar appearances for AI agents with high similarity of feature information. For example, cosine similarity can be used as the similarity of feature information. Also, the feature information of other characters to be used for calculating similarity should be generated and stored in the storage unit 11A or the like at the latest by the time the similarity is calculated.

[0060] (Content generation example 2: Generating new content based on the generated content) As explained with reference to Figure 4, the information processing device 1A can generate characters tailored to the company that has introduced the AI ​​agent. Furthermore, the information processing device 1A can also generate new characters based on the characters it has generated. This will be explained with reference to Figure 5. Figure 5 shows other examples of content generation by the information processing device 1A.

[0061] In the example in Figure 5, new characters C11 to C13 are generated based on character C1 shown in Figure 4. Specifically, in the example in Figure 5, prompt P3, which instructs the generation model M2 to generate new characters based on character C1, is input to the generation model M2 along with character C1. As a result, characters C11 to C13 are output from the generation model M2. The process of generating prompt P3 is performed by the prompt generation unit 101A, and the process of generating characters C11 to C13 using prompt P3 is performed by the content generation control unit 102A.

[0062] Furthermore, prompt P3 displays the "characteristics of the employee." Prompt P3 instructs the system to adapt the character for an employee who possesses those characteristics. Specifically, the employee characteristic shown in prompt P3 is "love of animals," which leads to the generation of characters C11-C13, whose designs incorporate animal elements into character C1.

[0063] In this way, the direction of the arrangement can also be specified via prompts. The information necessary to specify the direction of the arrangement to the generation model M2 (for example, information indicating the characteristics of employees using the AI ​​agent) can be obtained as related information by the data acquisition unit 103A. In addition, it is also possible to provide information indicating the characteristics of multiple companies as related information, and generate characters that are adapted from the base character for each of those companies.

[0064] Furthermore, prompt P3 instructs the system to create three different arranged characters. In this way, the prompt generation unit 101A and the content generation control unit 102A can also generate multiple types of content at once. It should be noted that the ability to generate multiple types of content at once is not limited to the generation of arranged content. For example, if prompt P2 in Figure 4 includes a statement instructing the system to generate multiple image data, it is possible to generate multiple image data at once.

[0065] Similarly, it is possible to generate multiple variations of a single piece of content. For example, if the previously generated content is image data of a character, it is possible to generate image data of that character with a different expression. In the same way, it is possible to generate image data of the character in different situations, such as when the character is busy working or when they have nothing to do. By using the information processing device 1A, it is possible to easily achieve such detailed differentiation of character depictions.

[0066] Furthermore, the image data of various expressions and situations created in this way can be utilized, for example, by displaying it according to the operating status of the object corresponding to that character. For example, if character image data of a certain AI agent is generated, it is possible to switch between image data showing the AI ​​agent working busily and image data showing it idle, depending on the situation in which the AI ​​agent is performing a task. To perform such processing, the information processing device 1A can be equipped with a monitoring unit to monitor the execution status of the AI ​​agent's task, and the presentation control unit 105A can determine the image data to present to the user according to the execution status identified by the monitoring unit.

[0067] As described above, the prompt generation unit 101A may generate a new prompt that instructs the generation model M2 to generate new content by arranging the base content, using the content generated by the generation model M2 as the base content. The content generation control unit 102A may then use the newly generated prompt to cause the generation model M2 to generate new content. With this configuration, in addition to the effects performed by the information processing device 1, the effect of easily generating content by arranging already generated content can be obtained.

[0068] Furthermore, the base content is not necessarily limited to content generated by the generation model M2. For example, base content generated by another device or created manually may be included in the related information. In this case, the prompt generation unit 101A generates a prompt instructing the system to generate content by referring to the base content included in the related information. With this configuration, in addition to the effects performed by the information processing device 1, the effect of easily generating content by arranging any content can be obtained.

[0069] The method for referencing the base content should be tailored to the generative model M2 being used. For example, if generative model M2 is capable of accepting both images and text as input, and the base content is image data, the base content should be input to generative model M2 along with a text-based prompt as shown in Figure 5. If generative model M2 is capable of accepting only text as input, a description of the base content should be generated and included in the prompt. The image data description can be generated by a generative model trained to generate such descriptions from image data. Alternatively, the user can be asked to input the image data description.

[0070] As described above, the information processing device 1A can generate individualized image data of multiple characters so that the functions of the AI ​​agent can be recognized from their appearance, and the differences between them and other characters corresponding to the same AI agent can also be recognized. When generating image data of multiple characters, it is also possible to specify via a prompt what aspects of the characters' image data should be generated that are common and what aspects are different.

[0071] For example, the prompt generation unit 101A may generate a prompt that instructs the generation of image data for one character (a character different from that of other AI agents) for each AI agent. Furthermore, if one AI agent is used by multiple users, the prompt generation unit 101A may generate a prompt that instructs the generation of image data in which the character has a different appearance for each user, such as through the character's clothing, belongings, accessories, or coloring. For example, the prompt generation unit 101A may generate a prompt that instructs the generation of an image of a character generated for a certain AI agent wearing a uniform bearing the logo of the company using that AI agent. This makes it possible to generate image data in which the corresponding AI agent can be identified from the character, and the company using that AI agent can be identified from the uniform the character is wearing.

[0072] (Content generation example 3: An example of performing object evaluation as a preprocessing step) Figure 6 shows yet another example of content generation by the information processing device 1A. In the example in Figure 6, image data of an AI agent's character is generated from the result of an AI agent performing a predetermined task. In other words, in the example in Figure 6, as in the example in Figure 4, the "object" is the AI ​​agent and the "content" is the image data. Also, in the example in Figure 6, the "related information" is information indicating the result of the AI ​​agent performing a predetermined task.

[0073] In the example in Figure 6, the execution results of the task shown in the related information are evaluated using the language model M1. This process is performed by the preprocessor 104A. Specifically, the preprocessor 104A generates a prompt P4 that includes the task execution results by the AI ​​agent and instructs the AI ​​agent to be evaluated by referring to those execution results, and inputs the generated prompt P4 to the language model M1. The prompt P4 can be generated by embedding the task execution results into a standard template.

[0074] Furthermore, prompt P4 also indicates the evaluation items and evaluation criteria. By specifying the evaluation items and evaluation criteria in this way, it becomes possible to standardize the evaluation items and evaluation criteria used for generating multiple pieces of content. Note that the evaluation items and evaluation criteria are arbitrary; for example, the user can specify the evaluation items and evaluation criteria. Also, one evaluation criterion may be set, or multiple criteria may be set. In addition, prompt P4 also shows the execution results of the same task by a comparison AI agent. By including such execution results, relative evaluation results can be output. In the example in Figure 6, by inputting prompt P4 into language model M1, language model M1 outputs response A2, which shows the evaluation result of the AI ​​agent.

[0075] Next, the prompt generation unit 101A generates a prompt using the evaluation results described above. In the example in Figure 6, prompt P5 is generated using answer A2. Prompt P5 is a prompt that instructs the generation of content (specifically, image data of the AI ​​agent character).

[0076] Furthermore, prompt P5 contains the evaluation results of the AI ​​agent. Therefore, by using prompt P5, it becomes possible to generate character image data that reflects the evaluation results of the AI ​​agent.

[0077] Furthermore, prompt P5, similar to prompt P2 shown in Figure 4, indicates the "company using" the AI ​​agent, and includes a statement instructing that the AI ​​agent be used by that company and that the character design be adapted for that company. This makes it possible to reflect the characteristics of the company using the AI ​​agent in the character's image data.

[0078] In the example shown in Figure 6, the content generation control unit 102A inputs prompt P5 to the generation model M2, which in turn outputs character C2. Character C2 reflects the high analytical ability shown in the evaluation results, featuring an intelligent design with glasses and including a truck that evokes the transportation company that uses the AI. According to the information processing device 1A, by simply providing the AI ​​agent's task execution results and the company using the AI ​​as related information, it is possible to generate such a character that reflects the AI ​​agent's task execution ability and the entity using the AI ​​agent.

[0079] As described above, the preprocessing unit 104A may use a language model M1 that has been trained on natural language to evaluate objects from related information using one or more predetermined evaluation criteria. The prompt generation unit 101A may also generate prompts using the evaluation results from the preprocessing unit 104A. With this configuration, in addition to the effects performed by the information processing device 1, the effect of easily generating content that reflects the evaluation results of predetermined evaluation criteria can be obtained.

[0080] (Example of content revision) The content generation control unit 102A may use the content generated by the generation model M2 as the final content, or it may use a modified version of the content generated by the generation model M2 as the final content. In the latter case, the user may be allowed to specify the modifications. Alternatively, the user may be allowed to determine the modifications by referring to other already generated content. This will be explained with reference to Figure 7. Figure 7 shows an example of a UI screen for accepting user modifications.

[0081] In the example UI screen in Figure 7, character C3 is displayed as "Character generated for XYZ Corporation." This character is content generated by generation model M2 under the control of content generation control unit 102A. In addition, the example UI screen in Figure 7 displays other characters generated by generation model M2 under the control of content generation control unit 102A as "Past generation examples." Furthermore, each character is associated with a description indicating what type of user the character was generated for, specifically a description such as "For sales" or "For research staff."

[0082] Furthermore, the UI screen example in Figure 7 displays the message, "Select the parts you want to import and press the confirm button," along with the characters. In other words, the UI screen in Figure 7 accepts the selection of a portion of already generated content and incorporates that portion into the content to be modified.

[0083] For example, as shown in Figure 7, if the reception unit 106A receives an operation to specify a part of a character that has already been generated by the cursor CUR, the presentation control unit 105A may display a preview image of a new character C31 in which the specified part has been incorporated into character C3. After this, if the reception unit 106A receives an operation to select the confirmation button, the content generation control unit 102A updates the generated character C3 to the new character C31.

[0084] The method of modifying the content is arbitrary and not limited to the examples above. For example, it may be possible to allow users to delete parts of the generated content, or to allow them to directly edit and modify the generated content. Furthermore, the user may be allowed to adjust the color, size, etc., of each part of the generated content via display objects such as sliders.

[0085] (Example of content regeneration) Furthermore, the information processing device 1A can also generate new content by incorporating the characteristics of other content into already generated content. This will be explained with reference to Figure 8. Figure 8 shows an example of generating new content by incorporating the characteristics of other content into already generated content.

[0086] More specifically, Image 1 shown in Figure 8 is an example of a UI screen for receiving content selection. The presentation control unit 105A may display such a UI screen, and the reception unit 106A may accept operations to specify content via such a UI screen.

[0087] In Img1, multiple generated characters are displayed on a coordinate plane where the vertical axis represents analytical capability and the horizontal axis represents response speed. Of the displayed characters, character C3 is the most recently generated character, while the other characters were generated before character C3.

[0088] In Img1, each character is displayed in a position corresponding to the evaluation result of the AI ​​agent that corresponds to that character. The evaluation result of the AI ​​agent can be generated by the preprocessor 104A, as explained, for example, with reference to Figure 6. If the AI ​​agent is evaluated using three or more evaluation criteria, the user may be allowed to select the evaluation criteria to be applied as the coordinate axis. Furthermore, the display position of the characters may be determined from perspectives other than the evaluation result of the AI ​​agent. For example, the tasks that the AI ​​agent can perform, the entity using the AI ​​agent, or the timing of the character's creation may be used as criteria, and characters with the same or similar tasks, entities, or creation timings may be placed in close positions.

[0089] Furthermore, for example, the display control unit 105A may display a list of characters used by the same entity, based on the evaluation results of the corresponding AI agent. This allows the user to recognize how characters used by the same entity are depicted differently according to their evaluation results.

[0090] Furthermore, the display control unit 105A may group and display characters that share the same user or corresponding AI agent, or characters with similar evaluation results. This allows the user to recognize how each character with common or similar parts is depicted differently.

[0091] In the example in Figure 8, the cursor CUR is used to enclose characters C3, C4, and C5. The reception unit 106A receives this operation as an operation to specify characters C4 and C5 as reference characters for character C3. A reference character is a character that possesses the characteristics that the user wants to incorporate into character C3.

[0092] When a reference character is specified, the prompt generation unit 101A generates a new prompt P6 that instructs character C3 to generate new content by incorporating the characteristics of the specified reference characters C4 and C5.

[0093] The content generation control unit 102A then inputs a new prompt P6 to the generation model M2 along with the target character C3 and reference characters C4 and C5, resulting in the output of a new character C32 from the generation model M2. The new character C32 retains the characteristics of character C3 while incorporating the characteristics of reference characters C4 and C5. The method for referencing content can be the same as when generating new content by referencing base content as explained based on Figure 5, and should be applied according to the generation model M2 being used.

[0094] As described above, the presentation control unit 105A presents the content generated by the generation model M2, and the reception unit 106A may accept an operation to specify the presented content. In this case, the prompt generation unit 101A may generate a new prompt that instructs the target content to generate new content by incorporating the features of the specified content. The content generation control unit 102A may then cause the generation model M2 to generate new content using the newly generated prompt. With this configuration, in addition to the effects performed by the information processing device 1, it becomes possible to easily generate new content that incorporates the features of the content specified by the user.

[0095] Furthermore, the user may be prompted to input a prompt instructing the generation of new content. The prompt generation unit 101A may also generate new prompts based on user input. With these configurations, it is possible to generate new content by incorporating the characteristics of the content in a way desired by the user. For example, the prompt generation unit 101A can also generate prompts instructing the user to make the new content similar in color, atmosphere, or shape of parts to a specified content.

[0096] (Process flow) The processing flow performed by the information processing device 1A will be explained with reference to Figure 9. Figure 9 is a flowchart showing the processing flow performed by the information processing device 1A. The flowchart in Figure 9 includes each process of the content generation method according to this exemplary embodiment.

[0097] In S11, the data acquisition unit 103A acquires relevant information related to an object that is the target of content generation and is programmed to perform a predetermined task. For example, the data acquisition unit 103A may acquire relevant information that includes information useful for content generation and information unnecessary for content generation, as described in the documentation regarding the object.

[0098] In S12, the preprocessor 104A generates a prompt instructing the language model M1 to extract information to be used for content generation, that is, information useful for content generation, from the related information acquired in S11. Then, in S13, the preprocessor 104A inputs the prompt generated in S12 to the language model M1 and causes it to extract information to be used for content generation.

[0099] Furthermore, if information usable for evaluating an object, such as the execution results of a task, is acquired as related information in S11, the preprocessor 104A may generate a prompt in S12 to instruct the evaluation of the object. Then, in S13, the preprocessor 104A may use this prompt to cause the language model M1 to output the evaluation result of the object.

[0100] In S14 (prompt generation process), the prompt generation unit 101A generates a prompt that instructs the system to generate content related to the object based on the relevant information acquired in S11. More specifically, the prompt generation unit 101A generates a prompt that includes the information extracted in S12 and instructs the system to generate content by referring to that information. In S14, the prompt generation unit 101A may also generate a prompt that includes the user of the object, as shown in prompt P2 in Figure 4. In that case, the data acquisition unit 103A should acquire information indicating the user of the object as relevant information in S11 or later.

[0101] In S15 (Content Generation Control Processing), the content generation control unit 102A uses the prompt generated in S14 to cause the generation model M2, which generates content corresponding to the input prompt, to generate content.

[0102] In S16, the presentation control unit 105A presents the content generated in S15 to the user. At this time, the presentation control unit 105A may also present other content that may be of reference to the user in addition to the content generated in S15, as shown in the example in Figure 7 or Figure 8. If it is not necessary for the user to confirm the content, the process in S16 may be omitted. In this case, the process in S17 is also omitted and the process proceeds to S18.

[0103] In S17, the reception unit 106A accepts modifications to the content presented in S16. For example, the reception unit 106A may accept modifications that incorporate a portion of other content, as shown in the example in Figure 7. If no modifications are made, the process in S17 is skipped and the process proceeds to S18.

[0104] In S18, the content generation control unit 102A records the content generated in S15 (or the content with the modifications reflected if modifications were accepted in S17) to a predetermined storage location such as the storage unit 11A, thereby terminating the process shown in Figure 9. If the user of the generated content has been determined at the time of S18, the information processing device 1A may perform a process to transfer ownership of the generated content to that user.

[0105] Furthermore, if the user of the object is determined after the completion of processing in S18, processing may be resumed from S11. In that case, in S11, the data acquisition unit 103A acquires relevant information indicating the user of the object. Then, processing in S12 and S13 is omitted, and in S14, the prompt generation unit 101A generates a prompt instructing the system to generate new content corresponding to the user, using the recorded content as base content. As a result, in S15, the content generation control unit 102A can generate new content corresponding to the user of the object.

[0106] Furthermore, suppose that in S16, the presentation control unit 105A presents other content along with the content generated in S15, and in S17, the reception unit 106A receives the specification of other content. In this case, the process returns to S11, and the data acquisition unit 103A may acquire the specified other content as new related information. After that, the processing in S12 and S13 is omitted and the processing in S14 is performed, in which the prompt generation unit 101A generates a new prompt instructing the target content to generate new content by incorporating the features of the other content shown in the new related information. As a result, in S15, the content generation control unit 102A can generate new content that incorporates the features of the other content.

[0107] [Variation] The entities executing each process described in the exemplary embodiments above are arbitrary and not limited to the examples given. For example, a system having the same functions as the information processing devices 1 and 1A can be constructed using multiple devices that can communicate with each other. Furthermore, the entities executing each process shown in the flowchart of Figure 9 may be a single device (which can also be called a processor) or multiple devices (which can also be called processors).

[0108] [Examples of implementation using software] Some or all of the functions of the information processing devices 1 and 1A (hereinafter also referred to as "the above devices") may be implemented by hardware such as integrated circuits (IC chips) or by software.

[0109] In the latter case, each of the above devices is implemented, for example, by a computer that executes instructions for a program, which is software that realizes each function. An example of such a computer (hereinafter referred to as computer C) is shown in Figure 10. Figure 10 is a block diagram showing the hardware configuration of computer C, which functions as each of the above devices.

[0110] Computer C comprises at least one processor C1 and at least one memory C2. Memory C2 stores a program (content generation program) P that causes computer C to operate as each of the above-mentioned devices. In computer C, processor C1 reads program P from memory C2 and executes it, thereby realizing each of the above-mentioned devices.

[0111] For processor C1, for example, a CPU (Central Processing Unit), GPU (Graphic Processing Unit), DSP (Digital Signal Processor), MPU (Micro Processing Unit), FPU (Floating Point Number Processing Unit), PPU (Physics Processing Unit), TPU (Tensor Processing Unit), quantum processor, microcontroller, or a combination thereof can be used. For memory C2, for example, flash memory, HDD (Hard Disk Drive), SSD (Solid State Drive), or a combination thereof can be used.

[0112] Computer C may also be equipped with RAM (Random Access Memory) for loading program P at runtime and for temporarily storing various data. Furthermore, computer C may be equipped with communication interfaces for sending and receiving data with other devices. Additionally, computer C may be equipped with input / output interfaces for connecting input / output devices such as keyboards, mice, displays, and printers.

[0113] Furthermore, program P can be recorded on a non-temporary, tangible recording medium M that is readable by computer C. Such a recording medium M could be, for example, tape, disk, card, semiconductor memory, or programmable logic circuitry. Computer C can acquire program P via such a recording medium M. Program P can also be transmitted via a transmission medium. Such a transmission medium could be, for example, a communication network or broadcast waves. Computer C can also acquire program P via such a transmission medium.

[0114] Furthermore, each of the above functions of each of the above devices may be implemented by a single processor in a single computer, by multiple processors in a single computer working together, or by multiple processors in each of multiple computers working together. In addition, the programs for implementing each of the above functions in each of the above devices may be stored in a single memory in a single computer, distributed and stored in multiple memories in a single computer, or distributed and stored in multiple memories in each of multiple computers.

[0115] [Additional Notes] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims.

[0116] (Note A1) An information processing device comprising: prompt generation means that generates a prompt instructing the generation of content relating to an object programmed to perform a predetermined task, based on relevant information relating to the object; and content generation control means that causes a generation model that generates content corresponding to an input prompt to generate content using the prompt.

[0117] (Appendix A2) The information processing apparatus according to Appendix A1, wherein the related information includes information indicating the task, and the prompt generating means generates a prompt that instructs to generate content corresponding to the task.

[0118] (Note A3) The information processing apparatus according to Appendix A1 or A2, wherein the related information includes information indicating the user of the object, and the prompt generating means generates a prompt instructing the user to generate content corresponding to the user.

[0119] (Note A4) The information processing device according to any one of the appendices A1 to A3, wherein the related information includes base content which is the basis for the content to be generated by the generation model, and the prompt generation means generates a prompt which instructs to generate content by referring to the base content.

[0120] (Note A5) The information processing apparatus according to any one of the appendices A1 to A4, wherein the prompt generation means generates a new prompt instructing the generation model to generate new content by arranging the base content, and the content generation control means causes the generation model to generate new content using the new prompt.

[0121] (Note A6) An information processing apparatus according to any one of Appendix A1 to A5, comprising a preprocessing means for extracting information to be used for generating the content from the related information using a language model that has been trained on natural language, and a prompt generation means for generating the prompt using the information extracted by the preprocessing means.

[0122] (Note A7) An information processing apparatus according to any one of the appendices A1 to A5, comprising a preprocessing means for evaluating the object from the related information using a machine learning language model for natural language according to one or more predetermined evaluation criteria, and a prompt generation means for generating the prompt using the evaluation results from the preprocessing means.

[0123] (Note A8) An information processing apparatus according to any one of Appendix A1 to A7, comprising: presentation control means for presenting content generated by the generation model; and reception means for receiving an operation to specify the presented content, wherein the prompt generation means generates a new prompt instructing the target content to generate new content by incorporating the characteristics of the specified content; and the content generation control means causes the generation model to generate new content using the new prompt.

[0124] (Note B1) A content generation method comprising: a prompt generation process in which at least one processor generates a prompt that instructs the generation of content relating to an object programmed to perform a predetermined task, based on relevant information relating to the object; and a content generation control process that uses the prompt to cause a generation model that generates content corresponding to the input prompt to generate content.

[0125] (Note B2) The content generation method according to Appendix B1, wherein the related information includes information indicating the task, and in the prompt generation process, the at least one processor generates a prompt instructing to generate content corresponding to the task.

[0126] (Note B3) The content generation method according to Appendix B1 or B2, wherein the related information includes information indicating the user of the object, and in the prompt generation process, the at least one processor generates a prompt instructing the user to generate content corresponding to the user.

[0127] (Note B4) The content generation method according to any one of Appendix B1 to B3, wherein the related information includes base content which is the basis for the content to be generated by the generation model, and in the prompt generation process, the at least one processor generates a prompt which instructs to generate content by referring to the base content.

[0128] (Note B5) The content generation method according to any one of the appendices B1 to B4, wherein the at least one processor generates a new prompt instructing the generator to generate new content by arranging the content generated by the generator model, using the content generated by the generator model as base content, and causes the generator model to generate new content using the new prompt.

[0129] (Note B6) A content generation method according to any one of Appendix B1 to B5, wherein the at least one processor includes preprocessing to extract information to be used for generating the content from the related information using a language model that has been trained on natural language, and in the prompt generation process, the at least one processor generates the prompt using the information extracted in the preprocessing.

[0130] (Note B7) The content generation method according to any one of the appendices B1 to B5, wherein the at least one processor includes preprocessing to evaluate the object from the related information using a language model that has been trained on natural language, according to one or more predetermined evaluation criteria, and in the prompt generation process, the at least one processor generates the prompt using the evaluation results obtained in the preprocessing.

[0131] (Note B8) A content generation method according to any one of Appendix B1 to B7, comprising: a presentation control process in which at least one processor presents content generated by the generation model; and a reception process in which at least one processor accepts an operation to specify the presented content, wherein the at least one processor generates a new prompt instructing the target content to generate new content by incorporating the features of the specified content, and causes the generation model to generate new content using the new prompt.

[0132] (Note C1) A content generation program that causes a computer to function as a prompt generation means that generates prompts instructing the computer to generate content related to an object programmed to perform a predetermined task, based on relevant information about the object, and a content generation control means that causes a generation model to generate content corresponding to the input prompt to generate content using the prompts.

[0133] (Note C2) The content generation program according to Appendix C1, wherein the related information includes information indicating the task, and the prompt generation means generates a prompt instructing to generate content corresponding to the task.

[0134] (Note C3) The content generation program described in Appendix C1 or C2, wherein the related information includes information indicating the user of the object, and the prompt generation means generates a prompt instructing the program to generate content corresponding to the user.

[0135] (Note C4) The content generation program described in any of Appendix C1 to C3, wherein the related information includes base content which is the basis for the content to be generated by the generation model, and the prompt generation means generates a prompt which instructs to generate content by referring to the base content.

[0136] (Note C5) The prompt generation means generates a new prompt that instructs the generation model to generate new content by arranging the base content, and the content generation control means causes the generation model to generate new content using the new prompt, according to any of the appendices C1 to C4.

[0137] (Appendix C6) A content generation program according to any one of the appendices C1 to C5, wherein the computer functions as a preprocessing means for extracting information to be used for generating the content from the related information using a language model that has been trained on natural language, and the prompt generation means generates the prompt using the information extracted by the preprocessing means.

[0138] (Note C7) The content generation program according to any one of the appendices C1 to C5, wherein the computer functions as a preprocessing means for evaluating the object from the related information using one or more predetermined evaluation criteria with respect to natural language using a language model that has been trained on natural language, and the prompt generation means generates the prompt using the evaluation results from the preprocessing means.

[0139] (Note C8) A content generation program as described in any of Appendix C1 to C7, wherein the computer functions as a presentation control means for presenting content generated by the generation model and a reception means for receiving an operation to specify the presented content, the prompt generation means generates a new prompt instructing the target content to generate new content by incorporating the characteristics of the specified content, and the content generation control means causes the generation model to generate new content using the new prompt.

[0140] (Note D1) An information processing device comprising at least one processor, wherein the at least one processor performs a prompt generation process that generates a prompt instructing to generate content relating to an object based on relevant information relating to an object programmed to perform a predetermined task, and a content generation control process that causes a generation model that generates content corresponding to an input prompt to generate content using the prompt.

[0141] The information processing device may also include memory. Furthermore, the memory may store a program that causes at least one processor to execute each of the aforementioned processes.

[0142] (Note D2) The information processing apparatus according to Appendix D1, wherein the related information includes information indicating the task, and in the prompt generation process, the at least one processor generates a prompt instructing to generate content corresponding to the task.

[0143] (Note D3) The information processing apparatus according to Appendix D1 or D2, wherein the related information includes information indicating the user of the object, and in the prompt generation process, the at least one processor generates a prompt instructing the user to generate content corresponding to the user.

[0144] (Note D4) The information processing apparatus according to any one of appendices D1 to D3, wherein the related information includes base content which is the basis for the content to be generated by the generation model, and in the prompt generation process, the at least one processor generates a prompt which instructs to generate content by referring to the base content.

[0145] (Note D5) The information processing apparatus according to any one of the appendices D1 to D4, wherein at least one processor generates a new prompt instructing the generation model to generate new content by arranging the base content, and causes the generation model to generate new content using the new prompt.

[0146] (Note D6) The information processing apparatus according to any one of the appendices D1 to D5, wherein at least one processor performs preprocessing to extract information to be used for generating the content from the related information using a language model that has been trained on natural language, and in the prompt generation process, the at least one processor generates the prompt using the information extracted in the preprocessing.

[0147] (Note D7) The information processing apparatus according to any one of the appendices D1 to D5, wherein at least one processor performs preprocessing to evaluate the object from the related information using a language model that has been trained on natural language, using one or more predetermined evaluation criteria, and in the prompt generation process, the at least one processor generates the prompt using the evaluation results obtained in the preprocessing process.

[0148] (Note D8) The information processing apparatus according to any one of the appendices D1 to D7, wherein the at least one processor performs a presentation control process for presenting content generated by the generation model and a reception process for receiving an operation to specify the presented content, the at least one processor generates a new prompt instructing the target content to generate new content by incorporating the features of the specified content, and causes the generation model to generate new content using the new prompt.

[0149] (Note E1) A non-temporary recording medium that records a content generation program, which includes a prompt generation process that generates a prompt instructing a computer to generate content related to an object programmed to perform a predetermined task, based on relevant information about the object, and a content generation control process that uses the prompt to cause a generation model that generates content corresponding to the input prompt to generate content. [Explanation of Symbols]

[0150] 1. Information Processing Device 101 Prompt generation unit (prompt generation means) 102 Content generation control unit (content generation control means) 1A Information Processing Device 101A Prompt generation unit (prompt generation means) 102A Content generation control unit (content generation control means) 104A Pre-processing unit (pre-processing means) 105A Presentation Control Unit (Presentation Control Means) 106A Reception area (reception method) M1 Generative Model M2 Language Model

Claims

1. A prompt generation means that generates a prompt instructing to generate content related to an object programmed to perform a predetermined task, based on relevant information about the object, An information processing device comprising: a content generation control means that causes a generation model, which generates content corresponding to an input prompt, to generate content using the aforementioned prompt.

2. The aforementioned related information includes information indicating the task, The information processing apparatus according to claim 1, wherein the prompt generation means generates a prompt that instructs to generate content corresponding to the task.

3. The aforementioned related information includes information indicating the user of the object, The information processing apparatus according to claim 1 or 2, wherein the prompt generation means generates a prompt that instructs the user to generate content corresponding to the user.

4. The aforementioned related information includes the base content that forms the basis of the content to be generated by the generation model, The information processing apparatus according to claim 1 or 2, wherein the prompt generation means generates a prompt that instructs to generate content by referring to the base content.

5. The prompt generation means generates a new prompt that instructs the generation of new content by arranging the base content, using the content generated by the generation model as the base content. The information processing apparatus according to claim 1 or 2, wherein the content generation control means causes the generation model to generate new content using the new prompt.

6. The system includes a preprocessing means that uses a language model trained on natural language to extract information used for generating the content from the related information, The information processing apparatus according to claim 1 or 2, wherein the prompt generation means generates the prompt using the information extracted by the preprocessing means.

7. The system includes a preprocessing means that uses a machine learning-based language model to evaluate the object from the related information using one or more predetermined evaluation criteria, The information processing apparatus according to claim 1 or 2, wherein the prompt generation means generates the prompt using the evaluation result from the preprocessing means.

8. A presentation control means for presenting content generated by the aforementioned generation model, The system includes a receiving means for receiving an operation to specify the presented content, The prompt generation means generates a new prompt that instructs the target content to generate new content by incorporating the specified characteristics of the content. The information processing apparatus according to claim 1 or 2, wherein the content generation control means causes the generation model to generate new content using the new prompt.

9. At least one processor, A prompt generation process that generates a prompt instructing the generation of content related to an object programmed to perform a predetermined task, based on relevant information about the object; A content generation method that performs a content generation control process that causes a generation model, which generates content according to the input prompt, to generate content using the aforementioned prompt.

10. Computers, A prompt generation means that generates a prompt instructing to generate content relating to an object based on relevant information relating to an object programmed to perform a predetermined task, and A content generation program that functions as a content generation control means, which uses the aforementioned prompt to cause a generation model that generates content in response to the input prompt to generate content.

Images