Device and method

Abstract

The present invention provides a device and a method that are capable of generating an appropriate prompt in accordance with generation target content to be generated and the purpose of generation.　In a prompt-generating device 100, an instruction acquisition unit 101, which is an instruction information acquisition unit, acquires instruction information relating to generation target content and the purpose of generation. A selection unit 102 selects, on the basis of the acquired instruction information, one prompt-generating AI 200 (for example, an image-specialized prompt-generating AI 200a) from among a plurality of prompt-generating AIs 200 for generating a response instruction prompt for issuing a response generation instruction to the generative AI 300. A prompt acquisition unit 105 acquires a response instruction prompt for a content generation instruction generated by the one selected prompt-generating AI 200. A response acquisition unit 106 transmits the response instruction prompt to the generative AI 300 to acquire a response result.

Description

Device and Method

[0001] The present invention relates to a device and a method using generative AI (Artificial Intelligence).

[0002] Patent Document 1 describes a prompt generation unit that generates a prompt by adding, as reference information, an additional sentence generated by an additional sentence generation unit to an input question sentence.

[0003] Japanese Patent No. 7325152

[0004] However, in the technology described in Patent Document 1, a prompt according to the purpose cannot be generated. Generally, in order to use generative AI, it is necessary for the user to manually create a prompt and try several times through trial and error. Also, depending on the information form (image, text, etc.) to be generated and the generation purpose, the way of writing the prompt is different, and the generative AI technology used is different.

[0005] Therefore, an object of the present invention is to provide a device and a method capable of generating an appropriate prompt according to the generation target content to be generated and the generation purpose.

[0006] The device of the present invention includes an instruction information acquisition unit that acquires instruction information regarding generation target content and its generation purpose, a selection unit that selects one prompt generation AI from a plurality of prompt generation AIs for generating an answer instruction prompt for giving an answer generation instruction to the generative AI based on the acquired instruction information, an answer instruction prompt acquisition unit that acquires an answer instruction prompt for content generation generated by the selected one prompt generation AI, and an answer result acquisition unit that transmits the answer instruction prompt to the generative AI and acquires the answer result thereof.

[0007] According to the present invention, an appropriate prompt can be generated according to the generation target content to be generated and the generation purpose.

[0008] Figure 1 is a diagram showing the system configuration including the prompt generation device 100 of this disclosure. Figure 2 is a schematic diagram showing the prompt generation process of this disclosure. Figure 3 is a diagram showing the input screen when input N3 is entered and the response instruction prompt generated based thereon. Figure 4 is a diagram showing a specific example of the text generation prompt P2 (response instruction prompt) generated based on input N3. Figure 5 is a block diagram showing the functional configuration of the prompt generation device of this disclosure. Figure 6 is a diagram showing the history information of evaluation results stored in the history storage unit 202. Figure 7 is a block diagram showing the functional configuration of the learning device 100a. Figure 8 is a flowchart showing the operation of the prompt generation device 100 to cause the prompt generation AI 200 and the generation AI 300 to generate content. Figure 9 is a flowchart showing the operation of the prompt generation device 100 to evaluate the response result. Figure 10 is a flowchart showing the operation of the prompt generation device 100 to predict the response result. Figure 11 is a diagram showing the transition from the instruction information input screen to the additional information request screen. Figure 12 is a diagram showing the evaluation reception screen on the user terminal 400 when an evaluation is accepted. Figure 13 shows other display screens on the user terminal 400. Figure 14 shows the configuration of the user terminal 400 on which the prompt generation device 100 or generation AI 200 / 300 is located. Figure 15 shows an example of the hardware configuration of the prompt generation device 100 and learning device 100a according to one embodiment of the present disclosure.

[0009] Embodiments of this disclosure will be described with reference to the attached drawings. Where possible, the same parts will be denoted by the same reference numerals, and redundant descriptions will be omitted.

[0010] Figure 1 shows a system configuration including the prompt generation device 100 of this disclosure. As shown in the figure, this system includes the prompt generation device 100, a prompt generation AI 200, and a generation AI 300. The user terminal 400 is a terminal operated by the user. In this disclosure, the prompt generation AI 200 and the generation AI 300 for content generation are arranged separately, but they may be the same generation AI.

[0011] The prompt generation device 100 selects a prompt generation AI 200 that corresponds to the generation purpose and the target object based on the generation purpose, the target object (the content to be generated), and supplementary information entered at the user terminal 400, and requests the prompt generation AI 200 to generate a prompt (a response instruction prompt) for giving a response instruction (an instruction to generate the target object) to the generation AI 300.

[0012] The prompt generator 100 obtains a response instruction prompt from the prompt generation AI 200. The prompt generator 100 transmits the obtained response instruction prompt to the selected generation AI 300 and obtains its response result.

[0013] The prompt generator 100 transmits the response result to the user terminal 400. The user at the user terminal 400 checks the response result and evaluates it.

[0014] With such a system, users can use appropriately generated prompts (response instruction prompts) according to the generation purpose and target object to have the generating AI 300 generate response results.

[0015] Figure 2 is a schematic diagram illustrating the prompt generation process of this disclosure. As shown in the figure, input N1 includes: Generation purpose: Generation of an advertising image for a new product that is likely to be clicked. Generation target: Image. Supplementary information: Chocolate... Input N1 is information entered by the user. In addition, the figure illustrates inputs N2 and N3. Their generation purposes are: Generation purpose: Generation of text for ideas for new product development..., Generation purpose: Generation of appeal text for a new product..., etc. Based on this input N1, etc., the prompt generation AI 200 generates a prompt.

[0016] In Figure 2(a), the image-specific prompt generation AI 200a is selected, and the image-specific prompt generation AI 200a generates an image generation prompt P1 (corresponding to an answer instruction prompt) based on the generation instruction prompt generated based on input N1. The image-specific generation AI 300a receives the image generation prompt P1 as input and generates the answer result K1, which is the target object (target content). As shown in Figure 2(b), inputs N2 and N3 are input to the text-specific prompt generation AI 200b, which is selected. The text-specific prompt generation AI 200b generates a text generation prompt P2 (corresponding to an answer instruction prompt), and the text-specific generation AI 300b, which receives this as input, generates the answer result K2.

[0017] Figure 3 shows the input screen S3 of the user terminal 400 when input N3 is entered, and the response instruction prompt generated based on it. Figure 3(a) shows the input screen S3 after input N3 has been entered. As shown in the figure, in the user terminal 400, the input screen S3 for input N3 accepts the generation purpose, the object to be generated, and supplementary information.

[0018] Here, the objects to be generated are selected from images, text, and music. The generation purpose and supplementary information are text entered by the user. In this disclosure, the supplementary information indicates the characteristics of the product. The supplementary information is supplementary information to achieve the generation purpose and specifies detailed information for generating a response instruction prompt. A generation start button is displayed on this input screen, and by selecting (clicking, tapping, etc.) this button, the user terminal 400 transmits this information to the prompt generation device 100.

[0019] In addition, this input screen S3 also includes Back, Home, Search Past Cases, and Help buttons, and the user can execute the respective function by tapping each button. Back returns to the previous screen, Home returns to the home screen, Search Past Cases proceeds to the past cases search screen, and Help proceeds to the help screen.

[0020] Figure 3(b) shows the generation instruction prompt input to the prompt generation AI 200. This generation instruction prompt is generated by the prompt generation device 100. As shown in the figure, this generation instruction prompt contains the command statement, "Please generate a prompt to cause the generation AI to create the 'text' of the 'new product appeal statement' using the following information." The command statement consists of the standard phrase "Please generate a prompt to cause the generation AI to create the '...' of '...' using the following information." The 'new product appeal statement' and 'text' are added to this standard phrase. The 'new product appeal statement' is information based on the generation purpose, and the 'text' is information based on the object to be generated. In addition, the generation purpose, object to be generated, and supplementary information based on input N3 are included in the generation instruction prompt. This standard phrase is stored in the prompt generation device 100.

[0021] Furthermore, this generation instruction prompt contains a description of the prompt description method selected in the generation AI selection model 103 (see Figure 5), which will be described later.

[0022] Figure 4 shows a specific example of a text generation prompt P2 (answer instruction prompt) generated based on input N3. As shown in the figure, the text generation prompt P2 includes the following as a standard phrase: "Please create multiple 'y's of 'x' based on the following conditions. Please let us know if additional information is needed. If so, please provide specific information about what kind of additional information it is." The standard phrase may be stored in the prompt generation AI 200, or it may be specified as a standard phrase in the generation instruction prompt.

[0023] In section P11, x is set to the generation purpose entered on the input screen. The main words or phrases analyzed by morphological analysis or similar methods in the text described for the purpose are set. Furthermore, in section P12 (the y portion), the generated object checked on the input screen is set. In this disclosure, text is set.

[0024] As a condition, section P13 is described as "approximately 10 characters" and "in a catchy slogan style." This section P13 is text generated by the prompt generation AI 200 and is information generated in response to the input information. In this disclosure, since it is a generated object, conditions related to characters are generated. Also, since it is a "new product appeal text," the prompt generation AI 200 generates conditions that it should be approximately 10 characters and in a catchy slogan style. In addition, supplementary information is set in section P14.

[0025] The generated text generation prompt P2 (response instruction prompt) is input to the generation AI 300, and the response result is generated. Here, the text for the new product appeal is generated by the generation AI 300 according to the specified conditions.

[0026] In this disclosure, the prompt generation device 100 generates prompts according to the prompt description method selected by the generation AI selection model. Figure 4 shows an example. Multiple prompt description methods are known, such as the Fukatsu method and the Junsuke method, depending on the purpose. Figure 4 shows a prompt generated using the Fukatsu method.

[0027] Figure 5 is a block diagram showing the functional configuration of the prompt generation device 100 of this disclosure. As shown in the figure, the prompt generation device 100 is composed of an instruction acquisition unit 101, a selection unit 102, a generation AI selection model 103, a prompt generation unit 104, a prompt acquisition unit 105, an answer acquisition unit 106, an evaluation acquisition unit 107, an evaluation prediction model 106a, and a history storage unit 202. The history storage unit 202 may be provided outside the prompt generation device 100.

[0028] The instruction acquisition unit 101 is the part that acquires instruction information from the user terminal 400. As described above, the instruction information includes the purpose of generation, the object to be generated, and supplementary information.

[0029] The selection unit 102 is responsible for selecting the prompt generation AI 200 and the generation AI 300 to be used based on the instruction information. Specifically, the selection unit 102 vectorizes the instruction information and inputs it into the generation AI selection model 103 (machine learning model). The selection unit 102 then obtains the output result from the generation AI selection model 103 and selects the prompt generation AI 200 / generation AI 300 and prompt description method to be used. As described above, the prompt description method indicates the format of the prompt, and an appropriate prompt description method is selected according to the object to be generated, the purpose of generation, etc. Furthermore, the vector conversion in this disclosure is performed using, for example, Word2Vec, GloVe, etc.

[0030] The Generative AI Selection Model 103 is a machine learning model that, upon input of vectorized instruction information, outputs information indicating a Generative AI that is suitable for that instruction information, as well as a prompt description method. The Generative AI Selection Model 103 is pre-trained using general machine learning methods, with instruction information for training (vectorized information) as explanatory variables and information indicating a Generative AI that is suitable for that instruction information, as well as a prompt description method, as the target variables. The information indicating a Generative AI is information that numerically identifies image Generative AI, text Generative AI, etc. In addition, several prompt description methods are defined to be output, and the model is trained so that the output of the most appropriate prompt description method is closer to 1.

[0031] The prompt generation unit 104 is the part that generates a generation instruction prompt for prompt generation instructions and sends it to the prompt generation AI 200. This generation instruction prompt is a prompt that causes the prompt generation AI 200 to generate a response instruction prompt based on the instruction information.

[0032] As shown in Figure 3(b), for example, a generation instruction prompt includes a command statement for generating a response instruction prompt to obtain a response result, the generation purpose described in the instruction information, the object to be generated, and supplementary information. This command statement is information that includes the intention of "generate a prompt to achieve the generation purpose," and includes information or text indicating the instruction to generate a prompt. Furthermore, this generation instruction prompt includes information indicating the use of the generation purpose, the object to be generated, and supplementary information.

[0033] The prompt acquisition unit 105 acquires a response instruction prompt from the prompt generation AI 200 (see Figure 4) and transmits it to the generation AI 300.

[0034] The response acquisition unit 106 acquires the response result, which is the response generated by the generation AI 300, and transmits the response result and an evaluation request for the response result to the user terminal 400. This evaluation request is screen information that allows the user to rate the response using the number of star marks, as will be described later (see Figure 12(b)).

[0035] Furthermore, the response acquisition unit 106 acquires the response result from the generating AI 300, and if the generating AI 300 generates additional information for the instruction information, or if any missing information is generated, it acquires that information. If the response acquisition unit 106 determines that additional information is necessary, it sends information to the user terminal 400 indicating that additional information is needed or that the specific additional information is needed, prompting the user to input the additional information and obtaining that additional information. The prompt generation unit 104 obtains the additional information from the response acquisition unit 106 and, taking the additional information into consideration, performs further generation instruction prompt generation processing.

[0036] Furthermore, when sufficient data is available, the response acquisition unit 106 determines whether the evaluation result is good or bad based on the evaluation result predicted from the evaluation prediction model 106a, according to the operator settings of the prompt generation device 100. If the response acquisition unit 106 determines that the evaluation result is bad, it then determines whether additional information is necessary based on the information regarding the necessity of additional information included in the response result.

[0037] The evaluation prediction model 106a is a predictive model that predicts the user's evaluation based on the instruction information and the response result. This evaluation prediction model 106a is a machine learning model and, as will be described later, is learned based on past history, which includes the instruction information, the response result, and the user's evaluation result at that time.

[0038] The evaluation acquisition unit 107 is responsible for acquiring the user's evaluation of the answer results transmitted to the user terminal 400 by the answer acquisition unit 106. The user evaluates the answer results displayed on the user terminal 400 by inputting a numerical value (for example, specifying the number of stars or clicking the star mark displayed on the screen of the user terminal 400). The evaluation acquisition unit 107 acquires that numerical value (number of stars) as the evaluation result. The evaluation acquisition unit 107 stores the acquired evaluation results, along with the instruction information and the answer results, in the history storage unit 202.

[0039] The history storage unit 202 is the part of the prompt generation device 100 that stores instruction information acquired by the instruction acquisition unit 101, response results acquired by the response acquisition unit 106, and evaluation results. This history storage unit 202 does not need to be included in the prompt generation device 100 and may be an external device. Figure 6 is a diagram showing the history information of evaluation results stored in the history storage unit 202. As shown in the figure, the history storage unit 202 stores instruction information, response results, and evaluation results in association with each other.

[0040] Next, the learning device 100a will be described. Figure 7 is a block diagram showing the functional configuration of the learning device 100a. As shown in the figure, the learning device 100a is composed of a learning unit 201 and a history storage unit 202. This learning device 100a is a device that uses the information stored in the history storage unit 202 to train an evaluation prediction model 106a for predicting evaluations of response results generated by the generating AI 300. The history storage unit 202 is the same as the one provided in the prompt generation device 100 and may be located outside the learning device 100a (in the prompt generation device 100).

[0041] The learning unit 201 is a part that learns the evaluation prediction model 106a by general machine learning using the instruction information and the answer result stored in the history storage unit 202 as explanatory variables and the evaluation result as the target variable. The instruction information and the answer result are vectorized and treated as explanatory variables. The evaluation result is treated as the number of stars as the target variable.

[0042] The history storage unit 202 is a part that stores the instruction information acquired by the instruction acquisition unit 101 of the prompt generation device 100, the answer result acquired by the answer acquisition unit 106, and the evaluation result.

[0043] Next, the operation of the prompt generation device 100 configured as described above will be described using FIGS. 8 to 10. FIG. 8 is a flowchart showing the operation of the prompt generation device 100 for causing the prompt generation AI 200 and the generation AI 300 to perform content generation. FIG. 9 is a flowchart showing the process of evaluating the answer result.

[0044] FIG. 8 is a flowchart showing the process in the case where the data is insufficient. For example, it shows the operation in the first stage where the number of evaluations of the answer result is not sufficient (for example, when the number of evaluations of the answer result is below the threshold). At the user terminal 400, input of instruction information including the generation purpose, the generation object, and supplementary information is performed. The instruction acquisition unit 101 acquires the instruction information from the user terminal 400 (S101). The selection unit 102 vector-converts the instruction information (S102). Then, the selection unit 102 inputs the converted vector to the generation AI selection model 103. The generation AI selection model 103 outputs the prompt generation AI 200 and the generation AI 300 as a selection result based on the vector (S103). Further, in the present disclosure, the prompt description method is also output.

[0045] The prompt generation unit 104 generates a generation instruction prompt for the selected prompt generation AI 200 based on the selection result and the prompt description method in order to cause the prompt generation AI 200 to generate a response instruction prompt (S104). The prompt generation unit 104 transmits the generation instruction prompt to the prompt generation AI 200. Then, the prompt acquisition unit 105 acquires the response instruction prompt (see FIG. 4) (S105).

[0046] The prompt acquisition unit 105 transmits the response instruction prompt to the generation AI 300. Then, the response acquisition unit 106 acquires the response result generated by the generation AI 300 (S106). This response result includes information indicating the necessity of additional information according to the judgment of the generation AI 300. When the response acquisition unit 106 determines that the response result includes information indicating that additional information is necessary (S107: necessary), it transmits an input request prompting the user terminal 400 to input additional information, and acquires the input additional information (S108).

[0047] The prompt generation unit 104 further generates a generation instruction prompt using the input additional information, and issues a generation request for a response instruction prompt to the prompt generation AI 200 (S104).

[0048] Also, as shown in FIG. 9, when it is determined in process S107 that additional information is unnecessary (S107: unnecessary), the response acquisition unit 106 transmits screen information indicating the response result to the user terminal 400 (S109).

[0049] On the user terminal 400, the user inputs an evaluation value for the response result, and the evaluation acquisition unit 107 acquires the evaluation value as an evaluation result and stores it in the history storage unit 202 together with the instruction information and the response result (S110).

[0050] Then, after an appropriate time, for example, when a predetermined number of history information is stored, the learning device 100a (learning unit 201) performs a learning process based on the instruction information and the evaluation result, and generates an evaluation prediction model 106a (S111).

[0051] The above process is for when the number of evaluation results is insufficient. Figure 10 is a flowchart showing the operation of the prompt generator 100 when there is sufficient data, for example, when the number of evaluation results is sufficient (the number of evaluation results is above a threshold). In this operation, the evaluation result is predicted from the instruction information and the response result based thereon, and appropriate processing such as requesting additional information is performed automatically based on that.

[0052] Processes S101 to S105 operate in the same way as in Figure 8. That is, the prompt generation device 100 obtains instruction information from the user terminal 400 and has the prompt generation AI 200 generate a response instruction prompt. The prompt generation device 100 then transmits the generated response instruction prompt to the generation AI 300 and has the generation AI 300 generate the response result.

[0053] The evaluation prediction model 106a predicts the evaluation result based on the instruction information and the response result (which may include information on whether additional information is needed) (S106a). The response acquisition unit 106 checks the evaluation result predicted by the evaluation prediction model 106a and determines whether the evaluation result is poor or good (S106b). If the evaluation result is poor and no additional information is needed, the process proceeds to S106, where the response acquisition unit 106 sends the same prompt again to the generating AI 300 and acquires the response result. If the evaluation result is poor and additional information is needed, the response acquisition unit 106 requests the user terminal 400 to input additional information and acquires it (S106c).

[0054] If the predicted evaluation result is good, the response acquisition unit 106 proceeds to process S109, where the response result screen information is transmitted, the user inputs the evaluation value, and the learning process is performed (S109 to S111).

[0055] Next, the display screen of the user terminal 400 will be described. Figure 11 shows the transition from the instruction information input screen to the additional information request screen. On the user terminal 400, the user inputs instruction information (see Figure 11(a), corresponding to S101 in Figure 8). In this disclosure, the user terminal 400 is, for example, a smartphone. The user inputs information from the input field for the generation purpose, the check field for the object to be generated, and the input field for supplementary information.

[0056] When the user terminal 400 transmits the input instruction information to the prompt generation device 100, the prompt generation device 100 generates a generation instruction prompt. Based on this generation instruction prompt, the prompt generation device 100 causes the prompt generation AI 200 to generate a response instruction prompt, and based on the response instruction prompt, causes the generation AI 300 to generate a response result. If the generation AI 300 determines that the response result is insufficient and additional information is needed, it adds to the response result a statement indicating that additional information is needed and provides specific information on what kind of additional information is required. The specific information is generated by the generation AI 300.

[0057] When the prompt generation device 100 (response acquisition unit 106) determines that the response result contains information indicating that additional information is needed and specific information about what kind of additional information is needed, it sends an input screen to the user terminal 400 to prompt the user to input the additional information (corresponding to S108 in Figure 8). This input screen includes information indicating that additional information is needed and specific information G.

[0058] Figure 11(b) shows the input screen for the additional information. Figure 11(b) is just one example, but according to this figure, the input screen includes the title "Additional information is needed" to indicate that additional information is required. The input screen also includes information G as specific additional information, such as "Please share product images and provide details about the appearance so that we can determine how well it will be received by children."

[0059] In Figure 11(b), the user attaches a product image T accordingly and sends it to the prompt generation device 100. While this disclosure includes attaching a product image as additional information, it is not limited to this; the user may also describe the product in text. Although omitted in Figure 11(b), this input screen includes a text input screen or an image file attachment screen. It also includes an input confirmation button for the user.

[0060] Through this exchange, the prompt generation device 100 can obtain additional information to generate more accurate response instruction prompts.

[0061] Figure 12 shows the evaluation reception screen on the user terminal 400 when an evaluation is received. Figure 12(a) shows the instruction information. By inputting the instruction information, the user can check the response result and provide an evaluation (corresponding to S109 in Figure 9).

[0062] Figure 12(b) shows the process of evaluating the response results. As shown in the figure, the response results generate new product appeal texts such as "A huge hit with kids! A new organic cake that's the talk of the town!" and "A chocolate cake topped with strawberries!". As shown in Figure 12(b), the evaluation screen displays the response result K and the evaluation result H. Multiple response results K are displayed on this evaluation screen, and the user can view each response result K and give an evaluation to each one. In this disclosure, the display screen is configured so that the user can select (click or tap, etc.) the part of the response result K. The user can evaluate each response result by adding a star mark. The number of star marks is controlled to increase according to the number of selections (number of clicks or taps). In the example in Figure 12(b), the upper limit is set to 3, but it is not limited to that.

[0063] Figure 12(c) shows the prompt display screen that appears when "Check prompt" is selected on the display screen of Figure 12(b). As shown in the figure, the response instruction prompt that yielded this answer result is shown here.

[0064] Furthermore, if "Regenerate" is selected on the display screen in Figure 12(b), the generation AI 300 generates the answer result using the same answer instruction prompt.

[0065] Figure 13 shows other display screens on the user terminal 400. This figure shows the display screen when "Search Past Cases" is selected from the instruction information input screen. Figure 13(a) shows the instruction information input screen. Figure 13(b) shows the display screen when "Search Past Cases" is selected. This figure shows the display screen for entering search keywords, and is a screen that can accept input of the target object and keywords. Figure 13(c) is the search results screen showing the search results. In Figure 13(c), the generation purpose, supplementary information, answer result, and evaluation result are shown for each case. By searching in this way, users can see the evaluation (number of star marks) of past answer results and refer to what kind of input information they should provide.

[0066] To perform this search process, the prompt generation device 100 includes a search unit, which can refer to the history storage unit 202 to retrieve corresponding answer results based on the object to be generated and the search keywords, and provide them to the user terminal 400.

[0067] Next, the effects of the prompt generation device 100 and the learning device 100a described herein will be explained.

[0068] In the prompt generation device 100 of this disclosure, the instruction acquisition unit 101, which is an instruction information acquisition unit, acquires instruction information relating to the content to be generated and the purpose of its generation. Based on the acquired instruction information, the selection unit 102 selects one prompt generation AI 200 (for example, an image-specific prompt generation AI 200a) from a plurality of prompt generation AIs 200 for generating a response instruction prompt to give a response generation instruction to the generation AI 300. The prompt acquisition unit 105 acquires a response instruction prompt for content generation instruction generated by the selected prompt generation AI 200. The response acquisition unit 106 transmits the response instruction prompt to the generation AI 300 and acquires the response result.

[0069] With this configuration, the prompt generation device 100 can select one prompt generation AI 200 from a plurality of prompt generation AIs 200 according to the instruction information. The instruction information includes the content to be generated and the purpose of generation, and therefore it is possible to generate an appropriate response instruction prompt according to the information format to be generated, and to obtain an appropriate response result based on it.

[0070] Furthermore, in this disclosure, the prompt generation unit 104 generates a prompt generation instruction prompt for prompt generation instruction based on instruction information, and transmits it to the selected prompt generation AI 200, for the purpose of causing the AI ​​200 to generate a response instruction prompt.

[0071] With this configuration, a generation instruction prompt for the prompt generation AI 200 is generated based on instruction information, thereby enabling the prompt generation AI 200 to generate an appropriate response instruction prompt.

[0072] Furthermore, multiple prompt generation AIs 200 are provided for each type of content to be generated. For example, there is an image-specific prompt generation AI 200a, a text-specific prompt generation AI 200b, and so on. The instruction information also includes information indicating the type of content to be generated (e.g., text, image, sound, etc.).

[0073] This enables the selection of instructions and prompt generation AI 200 according to the information format of the object to be generated.

[0074] Furthermore, the selection unit 102 selects a prompt generation AI 200 using the generation AI selection model 103. This generation AI selection model 103 is a machine learning model that has been trained with instruction information for learning as the explanatory variable and the prompt generation AI 200 determined according to the instruction information as the target variable.

[0075] This configuration allows for the selection of the appropriate prompt generation AI200.

[0076] Furthermore, if the response acquisition unit 106 determines that the response result contains information indicating that additional information is needed, it sends an input screen for the user (the source of the instruction information, such as the user terminal 400) to input the additional information. The prompt acquisition unit 105 then adds the additional information provided by the user and causes the prompt generation AI 200 to generate a response instruction prompt.

[0077] This configuration generates a generation instruction prompt that takes additional information into account, which is then sent to the prompt generation AI 200, which generates an appropriate response instruction prompt. Thus, an appropriate response can be obtained from the generation AI 300.

[0078] Furthermore, the prompt generation device 100 of this disclosure includes an evaluation prediction model 106a that outputs instruction information and a predicted evaluation result corresponding to the response result. The response acquisition unit 106, which functions as a reception unit, requests additional information from the user based on the predicted evaluation result output from the evaluation prediction model 106a and receives the additional information in accordance with the request. The prompt acquisition unit 105 then adds the additional information and causes a prompt generation AI 200 to generate a response instruction prompt.

[0079] With this configuration, if the prediction result evaluation is poor, a generation instruction prompt that takes additional information into account is generated, and this generation instruction prompt is sent to the prompt generation AI 200, which generates an appropriate response instruction prompt. Therefore, an appropriate response can be obtained from the generation AI 300. In addition, by predicting the user's evaluation before presenting the response result to the user, an appropriate response result can be obtained without causing any trouble to the user.

[0080] Furthermore, the response acquisition unit 106 of this disclosure transmits a request screen to the user for requesting an evaluation of the response result. The evaluation result given on the request screen is then stored in the history storage unit 202 in association with the response result and instruction information. The instruction information, response result, and evaluation result stored in the history storage unit 202 are used to train the evaluation prediction model 106a of the evaluation result in response to the input instruction information and response result.

[0081] Furthermore, the learning device 100a of this disclosure learns an evaluation prediction model 106a. The learning device 100a includes a history storage unit 202 that stores the answer result, instruction information, and evaluation result in association with each other. The learning unit 201 then learns the evaluation prediction model 106a using the instruction information and answer result stored in the history storage unit 202 as explanatory variables and the evaluation result as the objective variable.

[0082] This configuration allows the predictive model to learn based on the user's past instructions, responses, and evaluation results, thereby obtaining appropriate predictive and evaluation results.

[0083] The apparatus and method of this disclosure have the following configurations.

[0084] [1] An apparatus comprising: an instruction information acquisition unit that acquires instruction information relating to the content to be generated and the purpose of its generation; a selection unit that selects one prompt generation AI from a plurality of prompt generation AIs for generating an answer instruction prompt for giving an answer generation instruction to a generation AI based on the acquired instruction information; an answer instruction prompt acquisition unit that acquires an answer instruction prompt for content generation instructions generated by the selected one prompt generation AI; and an answer result acquisition unit that transmits the answer instruction prompt to the generation AI and acquires the answer result.

[0085] [2] The apparatus according to [1], further comprising: a prompt generation unit that generates a prompt generation instruction prompt for prompt generation instruction based on the instruction information for causing the selected prompt generation AI to generate the response instruction prompt; and transmits it to the selected prompt generation AI.

[0086] [3] The apparatus described in [1] or [2], wherein the plurality of prompt generation AIs are provided for each type of content to be generated.

[0087] [4] The apparatus according to any one of [1] to [3], wherein the instruction information includes information indicating the type of content to be generated.

[0088] [5] The apparatus according to any one of [1] to [4], wherein the selection unit selects the one prompt generation AI using a generation AI selection model, and the generation AI selection model is a machine learning model that has been trained with instruction information for learning as an explanatory variable and a prompt generation AI determined according to the instruction information as an objective variable.

[0089] [6] The apparatus according to any one of [1] to [5], wherein the response result acquisition unit determines that the response result includes information indicating that additional information is needed, and sends an input screen to the user for inputting additional information, and the response instruction prompt acquisition unit further adds the additional information and causes the prompt generation AI to generate the response instruction prompt.

[0090] [7] The apparatus according to any one of [1] to [5], further comprising: a prediction model that outputs a prediction evaluation result corresponding to the instruction information and the response result; and a receiving unit that requests additional information based on the prediction evaluation result output from the prediction model and receives additional information in accordance with the request, wherein the response instruction prompt acquisition unit further adds the additional information and causes the prompt generation AI to generate the response instruction prompt.

[0091] [8] The apparatus according to any one of [1] to [7], wherein the response result acquisition unit transmits a request screen to the user for requesting an evaluation of the response result, stores the evaluation result given on the request screen in a history storage unit in association with the response result and the instruction information, and the instruction information, the response result, and the evaluation result stored in the history storage unit are used to train a predictive model of the evaluation result in response to the input of the instruction information and the response result.

[0092] [9] [8] A learning device comprising: a history storage unit that stores the response result, the instruction information, and the evaluation result obtained from the device described in claim 8 in association with each other; and a learning unit that learns a predictive model using the instruction information and the response result stored in the history storage unit as explanatory variables and the evaluation result as the target variable.

[0093]

[10] A method for a device that communicates with multiple prompt generation AIs, comprising: an instruction information acquisition step of acquiring instruction information relating to content to be generated and the purpose of generating the content; a selection step of selecting one prompt generation AI from a plurality of prompt generation AIs for generating a response instruction prompt for giving a response generation instruction to the generation AI based on the acquired instruction information; a response instruction prompt acquisition step of acquiring a response instruction prompt for content generation instruction generated by the selected one prompt generation AI; and a response result acquisition step of transmitting the response instruction prompt to the generation AI and acquiring the response result.

[0094] Furthermore, all or some of the prompt generation AI 200 and generation AI 300, and the history storage unit 202 may be located in the prompt generation device 100 or the user terminal 400. Alternatively, the user terminal 400 may have the functions of the prompt generation device 100 and function as the prompt generation device 100. Some generation AI models, like Tsuzumi, have the generation AI model located within the user terminal 400. In this type, the RAG application is also provided on the user terminal 400. However, the information (knowledge DB) accessed by the RAG may be located within the user terminal 400 or on the network. There are also types, like ChatGPT, where the generation AI model is located on the network. In this type, the RAG application is provided on the user terminal 400. However, the information (knowledge DB) accessed by the RAG is on the network.

[0095] The specific configuration will now be explained. Figure 14(a) shows an example configuration when the user terminal 400 is equipped with the functions of the prompt generation device 100. In this case, the user terminal 400 can generate a prompt, send it to the generation AI 200 / 300, and obtain the result.

[0096] Figure 14(b) shows an example configuration when the user terminal 400 is equipped with a prompt generation device 100 and a generation AI 200 / 300. As shown in the figure, the user terminal 400 generates a prompt, outputs it to the built-in generation AI model 200 / 300, and obtains the result.

[0097] Incidentally, a generative AI model, such as the generative AI 200, is a model that, in response to a prompt containing input information, generates content according to the instructions, context, questions, and output format indicated by the prompt, and returns that content as response information. The prompt can also include input information, in which case the generative AI model generates response information targeting the input information. The generative AI model may be an interactive AI model that includes, for example, a Large Language Model (LLM) and a user interface (UI) for interaction with the user, enabling text chat or voice chat with the user. Examples of such generative AI models include ChatGPT, GPT®-3.5, GPT-4V, PaLM2, etc. In this embodiment, the prompt generation device 100 makes it possible to provide content provision functions using multiple types of interactive AI models. These interactive AI models may be stored within the prompt generation device 100, or they may be stored in other devices connected to the prompt generation device 100 via a network, and configured to exchange information with the user via the prompt generation device 100. Although only one prompt generation device 100 is shown in the figure, it may include multiple prompt generation devices 100.

[0098] In this disclosure, a prompt refers to information indicating instructions or questions entered by a user in an interactive system such as a dialogue with a generated AI model or a command-line interface (CLI).

[0099] The block diagram used in the description of the above embodiment shows functional units. These functional blocks (components) are realized by any combination of at least one of hardware and software. Furthermore, the method of realizing each functional block is not particularly limited. That is, each functional block may be realized using one device that is physically or logically coupled, or it may be realized using two or more physically or logically separated devices that are directly or indirectly connected (for example, using wired or wireless connections). A functional block may be realized by combining the one or more devices with software.

[0100] Functions include, but are not limited to, judgment, decision, determination, calculation, calculation, processing, derivation, investigation, exploration, confirmation, reception, transmission, output, access, resolution, selection, selection, establishment, comparison, assumption, expectation, assumption, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating (mapping), and assigning. For example, a functional block (configuration part) that enables transmission is called a transmitting unit or transmitter. In all cases, as mentioned above, the method of implementation is not particularly limited.

[0101] For example, the prompt generation device 100 and learning device 100a in one embodiment of the present disclosure may function as a computer that processes the prompt generation method of the present disclosure. Figure 15 is a diagram showing an example of the hardware configuration of the prompt generation device 100 and learning device 100a according to one embodiment of the present disclosure. The above-described prompt generation device 100 and learning device 100a may be physically configured as a computer device including a processor 1001, memory 1002, storage 1003, communication device 1004, input device 1005, output device 1006, bus 1007, etc.

[0102] In the following explanation, the term "device" can be replaced with "circuit," "device," "unit," etc. The hardware configuration of the prompt generation device 100 and the learning device 100a may include one or more of the devices shown in the figure, or it may be configured to omit some of the devices.

[0103] Each function in the prompt generation device 100 and the learning device 100a is realized by loading predetermined software (programs) onto hardware such as the processor 1001 and memory 1002, which causes the processor 1001 to perform calculations, control communication by the communication device 1004, and control at least one of data reading and writing in the memory 1002 and storage 1003.

[0104] The processor 1001 controls the entire computer, for example, by running the operating system. The processor 1001 may be composed of a central processing unit (CPU) that includes interfaces with peripheral devices, control devices, arithmetic units, registers, etc. For example, the selection unit 102 and prompt generation unit 104 described above may be implemented by the processor 1001.

[0105] Furthermore, the processor 1001 reads programs (program code), software modules, data, etc., from at least one of the storage 1003 and the communication device 1004 into the memory 1002 and executes various processes accordingly. The program used is one that causes the computer to execute at least a part of the operations described in the above embodiment. For example, the selection unit 102 and the prompt generation unit 104 may be stored in the memory 1002 and implemented by a control program that operates on the processor 1001, and other functional blocks may be implemented similarly. The above-described various processes have been explained as being executed by one processor 1001, but they may be executed simultaneously or sequentially by two or more processors 1001. The processor 1001 may be implemented by one or more chips. The program may also be transmitted from a network via a telecommunications line.

[0106] The memory 1002 is a computer-readable recording medium and may consist of at least one of the following: ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), RAM (Random Access Memory), etc. The memory 1002 may also be called a register, cache, main memory, etc. The memory 1002 can store executable programs (program code), software modules, etc., for carrying out the prompt generation method according to one embodiment of the present disclosure.

[0107] The storage 1003 is a computer-readable recording medium and may consist of at least one of the following: an optical disc such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, a magneto-optical disk (e.g., a compact disc, a digital multipurpose disc, a Blu-ray® disc), a smart card, flash memory (e.g., a card, a stick, a key drive), a floppy® disk, a magnetic strip, etc. The storage 1003 may also be called an auxiliary storage device. The above-mentioned storage medium may be, for example, a database, server, or other suitable medium including at least one of memory 1002 and storage 1003.

[0108] The communication device 1004 is hardware (transmitting / receiving device) for communicating between computers via at least one of a wired network and a wireless network, and is also referred to as a network device, network controller, network card, communication module, etc. The communication device 1004 may be configured to include high-frequency switches, duplexers, filters, frequency synthesizers, etc., in order to implement at least one of frequency division duplex (FDD) and time division duplex (TDD). For example, the instruction acquisition unit 101 and the response acquisition unit 106 described above may be implemented by the communication device 1004. The communication device 1004 may be implemented with physically or logically separated transmitting and receiving units.

[0109] The input device 1005 is an input device that accepts input from an external source (e.g., a keyboard, mouse, microphone, switch, button, sensor, etc.). The output device 1006 is an output device that outputs to an external source (e.g., a display, speaker, LED lamp, etc.). The input device 1005 and the output device 1006 may be configured as an integrated unit (e.g., a touch panel).

[0110] Furthermore, each device, such as the processor 1001 and the memory 1002, is connected by a bus 1007 for communicating information. The bus 1007 may be configured using a single bus, or different buses may be configured for each device.

[0111] Furthermore, the prompt generation device 100 and the learning device 100a may be configured to include hardware such as a microprocessor, a digital signal processor (DSP), an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), and an FPGA (Field Programmable Gate Array), and some or all of each functional block may be realized by such hardware. For example, the processor 1001 may be implemented using at least one of these hardware components.

[0112] Information notification is not limited to the embodiments described herein and may be carried out by other means. For example, information notification may be carried out by physical layer signaling (e.g., DCI (Downlink Control Information), UCI (Uplink Control Information)), upper layer signaling (e.g., RRC (Radio Resource Control) signaling, MAC (Medium Access Control) signaling, broadcast information (MIB (Master Information Block), SIB (System Information Block))), other signals, or combinations thereof. RRC signaling may also be called RRC messages, and may be, for example, RRC Connection Setup messages, RRC Connection Reconfiguration messages, etc.

[0113] The processing procedures, sequences, flowcharts, etc., of each aspect / embodiment described in this disclosure may be reordered, provided they do not contradict each other. For example, the methods described in this disclosure present various step elements using exemplary order and are not limited to the specific order presented.

[0114] Input and output information may be stored in a specific location (e.g., memory) or managed using a management table. Input and output information may be overwritten, updated, or appended to. Output information may be deleted. Input information may be transmitted to other devices.

[0115] The determination may be made by a value represented by one bit (0 or 1), by a boolean value (true or false), or by a numerical comparison (for example, a comparison with a predetermined value).

[0116] Each aspect / embodiment described in this disclosure may be used individually, in combination, or switched between as needed during implementation. Furthermore, notification of specific information (e.g., notification that "X is") is not limited to explicit notification, but may also be implicit (e.g., by not providing such notification).

[0117] Although the present disclosure has been described in detail above, it will be clear to those skilled in the art that the present disclosure is not limited to the embodiments described herein. The present disclosure can be implemented in modified and altered forms without departing from the intent and scope of the present disclosure as defined by the claims. Accordingly, the descriptions in the present disclosure are illustrative and not intended to be restrictive in any way.

[0118] Software should be broadly interpreted to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, execution threads, procedures, functions, and so on, whether they are called software, firmware, middleware, microcode, hardware description languages, or by any other name.

[0119] Furthermore, software, instructions, information, etc., may be transmitted and received via a transmission medium. For example, if software is transmitted from a website, server, or other remote source using at least one of wired technologies (such as coaxial cable, fiber optic cable, twisted pair, or digital subscriber line (DSL)) and wireless technologies (such as infrared or microwave), then at least one of these wired and wireless technologies is included in the definition of a transmission medium.

[0120] The information, signals, etc. described in this disclosure may be represented using any of the various different techniques. For example, the data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description may be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.

[0121] In addition, terms used in this disclosure and terms necessary for understanding this disclosure may be replaced with terms having the same or similar meanings. For example, at least one of the channel and symbol may be a signal (signaling). Also, a signal may be a message. Furthermore, a component carrier (CC) may be called a carrier frequency, cell, frequency carrier, etc.

[0122] Furthermore, the information, parameters, etc., described in this disclosure may be expressed using absolute values, relative values ​​from a given value, or other corresponding information. For example, wireless resources may be indicated by an index.

[0123] The names used for the parameters described above are not restrictive in any way. Furthermore, the formulas and other expressions using these parameters may differ from those expressly disclosed in this disclosure. Various channels (e.g., PUCCH, PDCCH, etc.) and information elements can be identified by any suitable name, and therefore, the various names assigned to these various channels and information elements are not restrictive in any way.

[0124] In this disclosure, terms such as "Mobile Station (MS)," "user terminal," "User Equipment (UE)," and "terminal" may be used interchangeably.

[0125] A mobile station may also be referred to by those skilled in the art as a subscriber station, mobile unit, subscriber unit, wireless unit, remote unit, mobile device, wireless device, wireless communication device, remote device, mobile subscriber station, access terminal, mobile terminal, wireless terminal, remote terminal, handset, user agent, mobile client, client, or some other appropriate term.

[0126] As used in this disclosure, the terms “determining” and “determining” may encompass a wide variety of actions. “Determining” may include, for example, judging, calculating, computing, processing, deriving, investigating, looking up, searching, or inquiring (e.g., searching in a table, database, or other data structure), or ascertaining. “Determining” may also include, for example, receiving (e.g., receiving information), transmitting (e.g., sending information), inputting, outputting, or accessing (e.g., accessing data in memory). Furthermore, "judgment" and "decision" can include considering something as having been "judged" or "decided" after resolving, selecting, choosing, establishing, comparing, etc. In other words, "judgment" and "decision" can include considering something as having been "judged" or "decided" after some action. Also, "judgment (decision)" can be reinterpreted as "assuming," "expecting," or "considering."

[0127] The terms “connected,” “coupled,” or any variation thereof, mean any direct or indirect connection or coupling between two or more elements, and may include the presence of one or more intermediate elements between two elements that are “connected” or “coupled” with each other. The coupling or connection between elements may be physical, logical, or a combination thereof. For example, “connection” may be reinterpreted as “access.” As used in this disclosure, two elements may be considered to be “connected” or “coupled” with each other using at least one of one or more wires, cables, and printed electrical connections, and, in some non-limiting and non-exclusive examples, electromagnetic energy having wavelengths in the radio frequency domain, microwave domain, and optical (both visible and invisible) domain.

[0128] In this disclosure, the phrase "based on" does not mean "based solely on" unless otherwise specified. In other words, the phrase "based on" means both "based solely on" and "based at least on."

[0129] Any reference to elements using designations such as “first,” “second,” etc., as used in this disclosure does not generally limit the quantity or order of those elements. These designations may be used in this disclosure as a convenient way to distinguish between two or more elements. Accordingly, references to first and second elements do not imply that only two elements may be employed, or that the first element must precede the second element in any way.

[0130] Where the terms “include,” “including,” and their variations are used in this disclosure, these terms are intended to be inclusive, as is the term “comprising.” Furthermore, the term “or” as used in this disclosure is not intended to be exclusive OR.

[0131] In this disclosure, if articles are added by translation, such as a, an, and the in English, this disclosure may include the fact that the noun following these articles is plural.

[0132] In this disclosure, the term "A and B are different" may mean "A and B are different from each other." The term may also mean "A and B are each different from C." Terms such as "separate" and "combine" may be interpreted similarly to "different."

[0133] 100...Prompt generation device, 101...Instruction acquisition unit, 102...Selection unit, 103...Generating AI selection model, 104...Prompt generation unit, 105...Prompt acquisition unit, 106...Answer acquisition unit, 107...Evaluation acquisition unit, 106a...Evaluation prediction model, 201...History storage unit, 202...History storage unit, 100a...Learning device, 200...Prompt generation AI, 300...Generating AI, 400...User terminal.

Claims

1. An apparatus comprising: an instruction information acquisition unit that acquires instruction information relating to the content to be generated and the purpose of its generation; a selection unit that selects one prompt generation AI from a plurality of prompt generation AIs for generating an answer instruction prompt for giving an answer generation instruction to a generation AI based on the acquired instruction information; an answer instruction prompt acquisition unit that acquires an answer instruction prompt for content generation instructions generated by the selected one prompt generation AI; and an answer result acquisition unit that transmits the answer instruction prompt to the generation AI and acquires the answer result.

2. The apparatus according to claim 1, further comprising: a prompt generation unit that generates a prompt generation instruction prompt for prompt generation instruction based on the instruction information, for causing the selected prompt generation AI to generate the response instruction prompt, and transmits it to the selected prompt generation AI.

3. The apparatus according to claim 1, wherein the plurality of prompt generation AIs are provided for each type of content to be generated.

4. The apparatus according to claim 1, wherein the instruction information includes information indicating the type of content to be generated.

5. The apparatus according to claim 1, wherein the selection unit selects the one prompt generation AI using a generation AI selection model, and the generation AI selection model is a machine learning model that has been trained with instruction information for learning as an explanatory variable and a prompt generation AI determined according to the instruction information as the target variable.

6. The apparatus according to claim 1, wherein the response result acquisition unit determines that the response result includes information indicating that additional information is needed, and transmits an input screen to the user for inputting additional information, and the response instruction prompt acquisition unit further adds the additional information and causes the first prompt generation AI to generate the response instruction prompt.

7. The apparatus according to claim 1, further comprising: a prediction model that outputs a prediction evaluation result corresponding to the instruction information and the response result; and a receiving unit that requests additional information based on the prediction evaluation result output from the prediction model and receives additional information in accordance with the request, wherein the response instruction prompt acquisition unit further adds the additional information and causes the one prompt generation AI to generate the response instruction prompt.

8. The apparatus according to claim 1, wherein the response result acquisition unit transmits a request screen to the user for requesting an evaluation of the response result, stores the evaluation result given on the request screen in a history storage unit in association with the response result and the instruction information, and the instruction information, the response result, and the evaluation result stored in the history storage unit are used to train a predictive model of the evaluation result in response to the input of instruction information and the response result.

9. A learning device comprising: a history storage unit that stores the response result, instruction information, and evaluation result obtained from the device described in claim 8 in association with each other; and a learning unit that trains a predictive model using the instruction information and response result stored in the history storage unit as explanatory variables and the evaluation result as the target variable.

10. A method for a device that communicates with multiple prompt generation AIs, comprising: an instruction information acquisition step of acquiring instruction information relating to content to be generated and the purpose of generating it; a selection step of selecting one prompt generation AI from a plurality of prompt generation AIs for generating a response instruction prompt for giving a response generation instruction to the generation AI based on the acquired instruction information; a response instruction prompt acquisition step of acquiring a response instruction prompt for content generation instruction generated by the selected one prompt generation AI; and a response result acquisition step of transmitting the response instruction prompt to the generation AI and acquiring the response result.

Images