Information processing apparatus, information processing method, and information processing program

Abstract

To provide an information processor, an information processing method and an information processing program for outputting a large amount of sentences with high accuracy while suppressing costs.SOLUTION: Generating a prompt including a result of executing the second generated AI by the first executing unit, the prompt being for causing the first generated AI to execute remaining pieces of text generation processing other than the predetermined number of pieces of text generation processing executed using the second generated AI, and causing the first generated AI to execute the remaining pieces of text generation processing using the generated prompt, the predetermined number of pieces of text generation processing being included in a plurality of pieces of text generation processing successively executed using the first generated AI, the second generated AI having a higher text generation accuracy than the first generated AI.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to an information processing device, an information processing method, and an information processing program. [Background technology]

[0002] In recent years, various companies have been developing and popularizing generative AI that generates new content using models trained on various data. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2024-60907 Summary of the Invention [Problem to be solved by the invention]

[0004] However, there are differences in the performance of generative AI from each company, and there is a wide range of performance, such as high-cost models with many variations in sentences for the input prompt, and low-cost models with few variations in sentences for the input prompt. For this reason, traditionally, when trying to output a large amount of highly accurate sentences using a model with many variations in sentences, costs increase.

[0005] The present application has been made in consideration of the above, and aims to provide an information processing device, an information processing method, and an information processing program that can output a large amount of highly accurate text while keeping costs down. [Means for solving the problem]

[0006] The information processing device of the present application comprises a first execution unit that executes a predetermined number of sentence generation processes among a plurality of sentence generation processes executed consecutively using a first generation AI using a second generation AI that has higher sentence generation accuracy than the first generation AI; a generation unit that generates a prompt including the execution result of the second generation AI by the first execution unit, for causing the first generation AI to execute the remaining sentence generation processes other than the predetermined number of sentence generation processes executed using the second generation AI; and a second execution unit that causes the first generation AI to execute the remaining sentence generation processes using the generated prompt. [Effects of the Invention]

[0007] According to one aspect of the embodiment, it is possible to suppress costs and output a large amount of highly accurate text. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is a diagram showing a process executed by an information processing apparatus according to an embodiment. [Figure 2] FIG. 2 is a diagram illustrating an example of the configuration of an information processing system according to the embodiment. [Figure 3] FIG. 3 is a diagram illustrating an example of the configuration of the information processing device according to the embodiment. [Figure 4] FIG. 4 is a diagram illustrating an example of model information. [Figure 5] FIG. 5 is a flowchart showing the processing procedure of the processing executed by the information processing apparatus according to the embodiment. [Figure 6] FIG. 6 is a diagram illustrating an example of a hardware configuration. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, an information processing device, an information processing method, and an information processing program according to the present application (hereinafter referred to as "embodiments") will be described in detail with reference to the drawings. Note that the information processing device, the information processing method, and the information processing program according to the present application are not limited to these embodiments. Furthermore, the same components in the following embodiments will be denoted by the same reference numerals, and duplicated descriptions will be omitted.

[0010] (Embodiment) First, a process executed by an information processing device according to an embodiment will be described with reference to Fig. 1. Fig. 1 is a diagram showing a process executed by an information processing device according to an embodiment. Fig. 1 shows an example of operation of an information processing system S including an information processing device 1 according to an embodiment.

[0011] 1, an information processing system S according to the embodiment includes an information processing device 1, a user terminal 100, and a service providing device 200. Although Fig. 1 shows an example in which the information processing device 1 and the service providing device 200 are configured as separate entities, the information processing device 1 and the service providing device 200 may also be configured as an integrated server device.

[0012] As shown in Figure 1, the information processing system S of the embodiment executes a predetermined number of sentence generation processes among multiple sentence generation processes executed consecutively using a first generation AI using a second generation AI that has higher sentence generation accuracy than the first generation AI, generates a prompt including the execution result of the second generation AI, and causes the first generation AI to execute the remaining sentence generation processes other than the predetermined number of sentence generation processes executed using the second generation AI, and causes the first generation AI to execute the remaining sentence generation processes using the generated prompt.

[0013] In the following, the low-accuracy, low-cost model will be referred to as the first generation AI, and the high-accuracy, high-cost model will be referred to as the second generation AI. Here, accuracy refers to, for example, the number of variations (number of characters, types of words used, etc.) in the sentences output by the generation AI when the same prompt is input, and whether or not the AI ​​has any weaknesses in the field of the task of generating the sentences (the field of writing style to be generated, multiple languages, specialized fields, computer languages).

[0014] In other words, when the same prompt is input, the low-accuracy first generation AI uses fewer characters and fewer types of words than the high-accuracy second generation AI, and is therefore weak in the field of the task. In other words, the second generation AI is a model that uses a larger number of characters and a larger variety of words than the first generation AI, and has no weaknesses in the field of the task.

[0015] Furthermore, even if the first generation AI has low accuracy, it is possible to induce it to output sentences with accuracy comparable to that of the second generation AI by setting detailed prompt conditions.

[0016] The cost here refers to the financial or time costs involved in training and operating the generation AI. For example, the cost of training a generation AI is the combined cost of the amount of data the generation AI needs to learn, the training period, and the computer performance required for training when developing the generation AI in-house. The low-precision first generation AI requires less data to learn, a shorter training period, and lower computer performance than the high-precision second generation AI, resulting in lower costs. Furthermore, for example, when using the generation AI as a subscription, the first generation AI has a lower monthly usage fee (lower cost) than the second generation AI. The cost may also include the electricity costs during operation. For example, the second generation AI has higher precision (larger amount of data to learn and more processing power for prompts) than the first generation AI, resulting in higher electricity costs (higher cost).

[0017] In the present disclosure, when a large amount of text generation processing is performed by a generation AI, a portion of the processing is performed by a highly accurate and high-cost second generation AI, and a prompt including the highly accurate execution result of the second generation AI as a model answer is input to the first generation AI. This allows costs to be reduced compared to when the entire large amount of text generation processing is performed by the second generation AI, because a portion of the processing can be performed by the first generation AI. Furthermore, compared to when the entire large amount of text generation processing is performed by the first generation AI, because the model answer is created by the second generation AI, the first generation AI can also generate highly accurate texts. In other words, the information processing device 1 according to the embodiment can output a large amount of highly accurate texts while suppressing costs.

[0018] Specifically, first, the information processing device 1 receives a processing request for a text generation task from the service providing device 200 (step S1). The text generation task may be one text generation task or multiple text generation tasks.

[0019] A single text generation task is, for example, the task of translating or summarizing a single book (such as a novel, legal text, medical document, etc.). Multiple text generation tasks are, for example, tasks in which a generation AI creates answers to a large number of questions on a Q&A site. In other words, a single text generation task is one in which there are no breaks within the task (continuous sentences), while a multiple text generation task is one in which there are breaks within the task (broken down by question).

[0020] Whether there is one or multiple sentence generation tasks may be specified, for example, by a requester (such as a user who has made a processing request to the service providing device 200), or the information processing device 1 may be configured to automatically determine this by analyzing the contents of the sentence generation task.

[0021] Next, the information processing device 1 divides the received text generation task into a plurality of processing units (step S2). Each processing unit corresponds to a document generation process described below. For example, in the case of one text generation task, the information processing device 1 divides it into a plurality of processing units according to a predetermined division condition. Specifically, the information processing device 1 detects divisions in the development of a book, such as chapters or headings, and divides the task so that each division becomes a processing unit. If divisions in the development do not exist (cannot be detected), the task may be divided, for example, into a predetermined number of characters (or number of lines).

[0022] Furthermore, in the case of multiple sentence generation tasks, the information processing device 1 divides each sentence generation task into a processing unit. For example, in the case of a Q&A site, the information processing device 1 divides each question into a processing unit.

[0023] In this way, by dividing the sentence generation task, the information processing device 1 can clearly distinguish between a processing unit for the first generation AI in the subsequent stage and a processing unit for the second generation AI.

[0024] Next, the information processing device 1 executes a sentence generation process for the first processing unit using the second generation AI (step S3). Specifically, the information processing device 1 generates a prompt (hereinafter, a prompt for the second generation AI) for causing the second generation AI to execute the sentence generation process for the processing unit that executes the sentence generation process first among the multiple processing units. For example, when the sentence generation process is summary (or) translation, the information processing device 1 generates a prompt for the second generation AI that includes a set of the text of the first processing unit and a command statement that commands summary (or translation). For example, the information processing device 1 generates a prompt for the second generation AI that says, "Please summarize the following sentence. ~ (corresponding to the sentence)."

[0025] Furthermore, when the sentence generation process is creating an answer to a question, the information processing device 1 generates a prompt for the second generation AI that sets the question text, which is the first processing unit, and a command statement that commands the creation of an answer. For example, the information processing device 1 generates a prompt for the second generation AI that says, "Please create an answer to the following question. ~ (corresponding to the question)."

[0026] Then, the information processing device 1 inputs the generated prompt to the second generation AI and acquires the sentences (summary sentences and answer sentences) generated by the sentence generation process executed by the second generation AI as the execution results. That is, in step S3, a model answer is generated as the execution result of the second generation AI.

[0027] In FIG. 1, a prompt for the second generation AI is generated only for the first processing unit. However, for example, multiple processing units may be grouped by similarity, and a prompt for the second generation AI may be generated for each group. For example, when the processing units are questions, the information processing device 1 groups questions with similar content. The information processing device 1 vectorizes each question using, for example, Word2Vec, and determines similarity using cosine similarity, for example.

[0028] Then, for each group, the information processing device 1 generates a second AI-generated prompt for any one (or two or more) processing units among the multiple processing units (questions) included in each group.

[0029] In this way, the information processing device 1 groups multiple processing units into similar processing units and generates a model answer using the second generation AI for each group, thereby generating a model answer for each trend in the content of the processing units, thereby preventing variation in the accuracy of the execution results of the generation AI for each content of the processing units across the entire sentence generation task.

[0030] Next, the information processing device 1 generates a prompt that includes the execution result of the second generation AI and causes the first generation AI to execute the sentence generation process of the second and subsequent processing units (step S4).

[0031] Specifically, the information processing device 1 generates a prompt (hereinafter referred to as the prompt for the first generation AI) that includes the prompt for the second generation AI generated in step S3, the execution result of the second generation AI (model answer), and a command statement that instructs the execution of sentence generation processing for the second and subsequent processing units.

[0032] For example, the information processing device 1 generates a prompt for the first generation AI such as "AA (corresponding to the model answer) is the model answer. Please summarize the following sentence based on this. BB (corresponding to the second processing unit)." When generating a prompt for the first generation AI corresponding to the third processing unit, the above "BB" is replaced with the text of the third processing unit.

[0033] Next, the information processing device 1 uses the generated prompt to have the first generation AI execute the sentence generation process for the second and subsequent processing units (step S5). Specifically, the information processing device 1 inputs the generated prompt for the first generation AI and the second and subsequent processing units to the first generation AI, thereby obtaining the execution results of the sentence generation process for the second and subsequent processing units from the first generation AI.

[0034] Next, the information processing device 1 outputs the execution results of the generated AI (first generated AI and second generated AI) to the service providing device 200 (step S6), and the service providing device 200 provides the received execution results of the generated AI to the user terminal 100 (step S7).

[0035] Specifically, the information processing device 1 outputs the execution result of the second generation AI in step S3 and the execution result of the first generation AI in step S5 to the service providing device 200. Note that the information processing device 1 may provide the execution result of the generation AI directly to the user terminal 100.

[0036] In this way, the information processing device 1 according to the embodiment has the second generation AI, which has high sentence generation accuracy, execute some of the sentence generation processes among a large number of sentence generation tasks. Then, the information processing device 1 inputs a prompt including the execution result of the sentence generation process of the second generation AI to the first generation AI, and has the first generation AI execute the remaining sentence generation processes. As a result, the first generation AI executes the sentence generation process by referring to the execution result of the second generation AI, and therefore the first generation AI can execute a large number of sentence generation processes at low cost while ensuring the sentence generation accuracy of the second generation AI. In other words, the information processing device 1 according to the embodiment can output a large number of sentences with high accuracy while suppressing costs.

[0037] Although the above example shows a case where the sentence generation task is a sentence (text), the sentence generation task may be a voice or an image.

[0038] For example, in the case of speech, the information processing device 1 uses a large amount of speech data (which may be segmented or may be a single piece of data such as a long speech) as a sentence generation task, generates prompts for the first generation AI and the second generation AI in the same manner as described above, and generates a sentence based on the speech using the first generation AI and the second generation AI. Examples of sentences based on speech include sentences in which speech is directly converted into text, such as transcription, and sentences that summarize (or translate) the content of the speech.

[0039] For example, in the case of an image, the information processing device 1 takes a large amount of image data (which may be divided into sections or may be a single piece of data such as a long video) as a sentence generation task, generates prompts for the first generation AI and the second generation AI in the same manner as described above, and generates sentences based on the image using the first generation AI and the second generation AI. Examples of sentences based on the image include sentences that explain the content of the image and sentences that convert objects shown in the image into text.

[0040] Next, a configuration example of an information processing system S according to an embodiment will be described with reference to Fig. 2. Fig. 2 is a block diagram showing a configuration example of the information processing system S according to an embodiment. As shown in Fig. 2, in the information processing system S according to an embodiment, an information processing device 1, a plurality of user terminals 100, and a plurality of service providing devices 200 are connected to a network N by wire or wirelessly. The network N is, for example, a network such as the Internet, a WAN (Wide Area Network), or a LAN (Local Area Network).

[0041] The information processing device 1 is a server device that executes an information processing method according to an embodiment. The information processing device 1 executes a predetermined number of sentence generation processes, among a plurality of sentence generation processes executed consecutively using a first generation AI, using a second generation AI that has higher sentence generation accuracy than the first generation AI, generates a prompt including the execution result of the second generation AI, and causes the first generation AI to execute the remaining sentence generation processes other than the predetermined number of sentence generation processes executed using the second generation AI, and causes the first generation AI to execute the remaining sentence generation processes using the generated prompt.

[0042] Furthermore, the information processing device 1 is an information processing device that cooperates with multiple user terminals 100 and multiple service providing devices 200, and provides API (Application Programming Interface) services for various applications (hereinafter referred to as apps), etc., and various data to the multiple user terminals 100 and the multiple service providing devices 200, and is realized by a server device, a cloud system, etc.

[0043] Furthermore, the information processing device 1 may be an information processing device that provides some kind of online web service to a plurality of user terminals 100 and a plurality of service providing devices 200. For example, the information processing device 1 may provide services such as internet connection, search service, SNS (Social Networking Service), electronic commerce (EC), electronic payment, online games, online banking, online trading, hotel and ticket reservations, video and music distribution, news, maps, route search, route guidance, line information, operation information, and weather forecasts as web services. In practice, the information processing device 1 may cooperate with various servers that provide the above-mentioned web services and act as an intermediary for the web services or may be responsible for processing the web services.

[0044] The user terminal 100 is a terminal device owned by a user who uses a service provided by the service providing device 200. The user terminal 100 can be any type of terminal device, such as a smartphone, a desktop PC, a notebook PC, or a tablet PC. The user terminal 100 transmits various types of information to the information processing device 1 or the like, and receives information provided by the information processing device 1 or the like.

[0045] The service providing device 200 is a server device that provides various services to a user who owns the user terminal 100. The service providing device 200 transmits various information to the information processing device 1, the user terminal 100, etc., and receives information provided from the information processing device 1, the user terminal 100, etc.

[0046] Next, an example of the configuration of the information processing device 1 will be described with reference to FIG.

[0047] Fig. 3 is a diagram showing an example of the configuration of an information processing device 1 according to an embodiment. As shown in Fig. 3, the information processing device 1 has a communication unit 2, a control unit 3, and a storage unit 4. The control unit 3 includes a reception unit 31, a division unit 32, a first execution unit 33, a generation unit 34, a second execution unit 35, and a provision unit 36. The storage unit 4 stores model information 41.

[0048] The communication unit 2 is realized by, for example, a network interface card (NIC), etc. The communication unit 2 is connected to a network by wire or wirelessly.

[0049] The control unit 3 is a controller, and is realized by a processor such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit) executing various programs (corresponding to an example of an information processing program) stored in a storage device inside the information processing device 1 using a RAM or the like as a work area. The control unit 3 is also a controller, and may be realized by an integrated circuit such as an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a GPGPU (General Purpose Graphic Processing Unit).

[0050] The storage unit 4 is realized by, for example, a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk.

[0051] The model information 41 is information relating to the model of the generating AI.

[0052] Fig. 4 is a diagram showing an example of the model information 41. As shown in Fig. 4, the model information 41 includes items such as "model ID", "name", "performance", and "model parameters".

[0053] "Model ID" is identification information that identifies the model. "Name" is information that indicates the name of the model of the generating AI. "Performance" is information that indicates the performance of the model. "Model parameters" is information about the parameters of the model, and includes information such as neural networks and weighting coefficients in deep learning.

[0054] Next, each function of the control unit 3 of the information processing device 1 (receiving unit 31, dividing unit 32, first execution unit 33, generating unit 34, second execution unit 35, and providing unit 36) will be described.

[0055] The receiving unit 31 receives a processing request for a text generation task from the service providing device 200. The text generation task may be one text generation task or multiple text generation tasks.

[0056] The division unit 32 divides the received text generation task into a plurality of processing units. Each processing unit corresponds to a document generation process described below. For example, in the case of one text generation task, the division unit 32 divides the task into a plurality of processing units according to a predetermined division condition. Specifically, the division unit 32 detects divisions in the development of a book, such as chapters or headings, and divides the task so that each division becomes a processing unit. If divisions in the development do not exist (cannot be detected), the task may be divided, for example, into a predetermined number of characters (or number of lines).

[0057] Furthermore, in the case of a plurality of sentence generation tasks, the division unit 32 divides each sentence generation task into a processing unit. For example, in the case of a Q&A site, the division unit 32 divides each question into a processing unit.

[0058] The first execution unit 33 executes a sentence generation process using the second generation AI for the first processing unit. Specifically, the first execution unit 33 generates a prompt for the second generation AI that causes the second generation AI to execute a sentence generation process for the processing unit that is to execute the sentence generation process first among the multiple processing units. For example, if the sentence generation process is summarization (or) translation, the first execution unit 33 generates a prompt for the second generation AI that includes a set of the text of the first processing unit and a command statement that commands summarization (or translation). For example, the first execution unit 33 generates a prompt for the second generation AI that says, "Please summarize the following sentence. ~ (equivalent to the sentence)."

[0059] Furthermore, when the sentence generation process is creating an answer to a question, the first execution unit 33 generates a prompt for the second generation AI that sets the text of the question, which is the first processing unit, and a command statement that commands the creation of an answer. For example, the first execution unit 33 generates a prompt for the second generation AI that says, "Please create an answer to the following question. ~ (corresponding to the question)."

[0060] Then, the first execution unit 33 inputs the generated prompt to the second generation AI, and acquires the sentences (summary sentences and answer sentences) generated by the sentence generation process executed by the second generation AI as execution results.

[0061] Although the first execution unit 33 generated a prompt for the second generation AI only for the first processing unit, it may also be possible to group multiple processing units by similarity and generate a prompt for the second generation AI for each group. For example, if the processing units are questions, the first execution unit 33 groups questions with similar content. The first execution unit 33 vectorizes each question using, for example, Word2Vec, and determines similarity using cosine similarity, for example.

[0062] Then, for each group, the first execution unit 33 generates a second AI-generated prompt for any one (or two or more) processing units out of the multiple processing units (questions) included in each group.

[0063] The generation unit 34 generates a prompt that includes the execution result of the second generation AI and causes the first generation AI to execute the sentence generation process of the second and subsequent processing units.

[0064] Specifically, the generation unit 34 generates a prompt for the first generation AI that includes the prompt for the second generation AI generated by the first execution unit 33, the execution result of the second generation AI (model answer), and a command statement that commands the execution of sentence generation processing for the second and subsequent processing units.

[0065] For example, the generation unit 34 generates a prompt for the first generation AI such as "AA (corresponding to the model answer) is the model answer. Please summarize the following sentence based on this. BB (corresponding to the second processing unit)." When generating a prompt for the first generation AI corresponding to the third processing unit, the above "BB" is replaced with the text of the third processing unit.

[0066] The second execution unit 35 causes the first generation AI to execute the sentence generation process for the second and subsequent processing units using the generated prompt. Specifically, the second execution unit 35 inputs the generated prompt for the first generation AI and the second and subsequent processing units to the first generation AI, thereby obtaining the execution results of the sentence generation process for the second and subsequent processing units from the first generation AI.

[0067] The providing unit 36 ​​provides the execution results of the generation AIs (first generation AI and second generation AI) to the service providing device 200. Specifically, the providing unit 36 ​​outputs the execution results of the second generation AI and the execution results of the first generation AI to the service providing device 200. Note that the providing unit 36 ​​may also provide the execution results of the generation AIs directly to the user terminal 100.

[0068] Next, a processing procedure of the processing executed by the information processing device 1 according to the embodiment will be described with reference to Fig. 5. Fig. 5 is a flowchart showing the processing procedure of the processing executed by the information processing device 1 according to the embodiment.

[0069] As shown in FIG. 5, the control unit 3 first receives a processing request for a text generation task from the service providing device 200 (step S101).

[0070] Next, the control unit 3 divides the received text generation task into N processing units (step S102).

[0071] Next, the control unit 3 executes a sentence generation process for the first processing unit using the second generation AI (step S103).

[0072] Next, the control unit 3 generates a prompt for the Xth (X≧2) processing unit out of N (step S104).

[0073] Next, the control unit 3 executes the document generation process of the Xth processing unit using the first generation AI (step S105).

[0074] Next, the control unit 3 determines whether X is N, that is, whether the sentence generation process for the last N-th processing unit out of N has been completed (step S106).

[0075] If X is N, that is, if the sentence generation process for the last Nth processing unit out of N has been completed (step S106: Yes), the control unit 3 provides the execution result of the generation AI to the service providing device 200 (step S107) and terminates the process.

[0076] If X is not N, that is, if the sentence generation process for the last N-th processing unit among N has not been completed (step S106: No), the control unit 3 counts up X and returns to step S104.

[0077] 〔others〕 Furthermore, among the processes described in the above embodiments, some of the processes described as being performed automatically can also be performed manually. Alternatively, all or some of the processes described as being performed manually can be performed automatically using known methods. In addition, the information including the processing procedures, specific names, various data, and parameters shown in the above documents and drawings can be changed as desired unless otherwise specified. For example, the various information shown in each drawing is not limited to the information shown in the drawings.

[0078] Furthermore, the components of each device shown in the figure are conceptual functional components and do not necessarily have to be physically configured as shown in the figure. In other words, the specific form of distribution and integration of each device is not limited to that shown in the figure, and all or part of them can be functionally or physically distributed and integrated in any unit depending on various loads, usage conditions, etc.

[0079] 3 may be held in a storage server or the like, rather than being held by each device. In this case, each device obtains various pieces of information by accessing the storage server.

[0080] [Hardware configuration] The information processing device 1 according to the embodiment described above is realized by, for example, a computer 1000 configured as shown in Fig. 6. Fig. 6 is a diagram showing an example of a hardware configuration. The computer 1000 is connected to an output device 1010 and an input device 1020, and has a configuration in which a calculation device 1030, a primary storage device 1040, a secondary storage device 1050, an output IF (Interface) 1060, an input IF 1070, and a network IF 1080 are connected via a bus 1090.

[0081] The arithmetic device 1030 operates based on programs stored in the primary storage device 1040 and secondary storage device 1050, programs read from the input device 1020, and the like, and executes various processes. The primary storage device 1040 is a memory device, such as a RAM, that temporarily stores data used by the arithmetic device 1030 for various calculations. The secondary storage device 1050 is a storage device in which data used by the arithmetic device 1030 for various calculations and various databases are registered, and is realized by a ROM (Read Only Memory), an HDD (Hard Disk Drive), a flash memory, or the like.

[0082] The output IF 1060 is an interface for transmitting information to be output to an output device 1010 that outputs various types of information, such as a monitor or a printer, and is realized by a connector conforming to a standard such as USB (Universal Serial Bus), DVI (Digital Visual Interface), or HDMI (High Definition Multimedia Interface), etc. The input IF 1070 is an interface for receiving information from various input devices 1020, such as a mouse, keyboard, scanner, etc., and is realized by a USB, etc.

[0083] The input device 1020 may be a device that reads information from, for example, an optical recording medium such as a CD (Compact Disc), a DVD (Digital Versatile Disc), or a PD (Phase Change Rewritable Disk), a magneto-optical recording medium such as an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory. The input device 1020 may also be an external storage medium such as a USB memory.

[0084] The network IF 1080 receives data from other devices via the network N and sends it to the arithmetic device 1030, and also transmits data generated by the arithmetic device 1030 to other devices via the network N.

[0085] The arithmetic unit 1030 controls the output device 1010 and the input device 1020 via the output IF 1060 and the input IF 1070. For example, the arithmetic unit 1030 loads a program from the input device 1020 or the secondary storage device 1050 onto the primary storage device 1040 and executes the loaded program.

[0086] For example, when the computer 1000 functions as the information processing device 1, the arithmetic unit 1030 of the computer 1000 executes a program loaded onto the primary storage device 1040, thereby realizing the functions of the control unit 3.

[0087] 〔effect〕 As described above, the information processing device 1 according to the embodiment includes a first execution unit 33 that executes a predetermined number of sentence generation processes using a second generation AI that has higher sentence generation accuracy than the first generation AI among a plurality of sentence generation processes that are executed consecutively using a first generation AI, a generation unit 34 that generates a prompt including the execution result of the second generation AI by the first execution unit 33, the prompt being used to cause the first generation AI to execute the remaining sentence generation processes other than the predetermined number of sentence generation processes executed using the second generation AI, and a second execution unit 35 that causes the first generation AI to execute the remaining sentence generation processes using the generated prompt.

[0088] This configuration makes it possible to output a large amount of highly accurate text while keeping costs down.

[0089] Although some of the embodiments of the present application have been described in detail above with reference to the drawings, these are merely examples, and the present invention can be implemented in other forms that include the embodiments described in the Disclosure of the Invention section and that have undergone various modifications and improvements based on the knowledge of those skilled in the art.

[0090] 〔others〕 Furthermore, among the processes described in the above embodiments, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically using known methods. In addition, the information including the processing procedures, specific names, various data, and parameters shown in the above documents and drawings can be changed as desired unless otherwise specified. For example, the various information shown in each drawing is not limited to the information shown in the drawings.

[0091] Furthermore, the components of each device shown in the figure are conceptual functional components and do not necessarily have to be physically configured as shown in the figure. In other words, the specific form of distribution and integration of each device is not limited to that shown in the figure, and all or part of them can be functionally or physically distributed and integrated in any unit depending on various loads, usage conditions, etc.

[0092] Furthermore, the processes described in the above-described embodiments can be combined as appropriate within the scope of not causing any contradiction in the process contents.

[0093] Furthermore, the above-mentioned "section, module, unit" can be read as "means" or "circuit," etc. For example, the control unit 3 can be read as control means or a control circuit. [Explanation of symbols]

[0094] 1. Information processing equipment 2. Communications Department 3. Control Unit 4 Storage section 31 Reception 32 Division 33 First Executive Division 34 Generation part 35 Second Executive Division 36 Providing Department 41 Model Information 100 user terminals 200 Service Providing Device S Information Processing System

Claims

1. a first execution unit that executes a predetermined number of sentence generation processes among a plurality of sentence generation processes executed successively using a first generation AI using a second generation AI having a higher sentence generation accuracy than the first generation AI; a generation unit that generates a prompt including the execution result of the second generation AI by the first execution unit, the prompt causing the first generation AI to execute the remaining sentence generation processes other than the predetermined number of sentence generation processes executed using the second generation AI; a second execution unit that causes the first generation AI to execute the remaining sentence generation process using the generated prompt; An information processing device comprising:

2. a dividing unit that divides a sentence generation task, the processing request of which has been received, into a plurality of processing units; The plurality of sentence generation processes include: A process corresponding to each of the plurality of processing units The information processing device according to claim 1 .

3. The sentence generation task is a single task, The dividing unit Dividing one of the sentence generation tasks into the plurality of processing units The information processing device according to claim 2 .

4. the sentence generation task is a plurality of tasks, The dividing unit Dividing each of the plurality of sentence generation tasks into the plurality of processing units. The information processing device according to claim 2 .

5. The dividing unit grouping the plurality of processing units into groups of similar processing units; The first execution unit The sentence generation process is executed using the second generation AI for at least one of the processing units included in the group. The information processing device according to claim 2 .

6. 1. A computer-implemented information processing method, comprising: a first execution step of executing a predetermined number of sentence generation processes among a plurality of sentence generation processes executed successively using a first generation AI using a second generation AI having a higher sentence generation accuracy than the first generation AI; a generation step of generating a prompt including the execution result of the second generation AI by the first execution step, the prompt causing the first generation AI to execute the remaining sentence generation processes other than the predetermined number of sentence generation processes executed using the second generation AI; a second execution step of causing the first generation AI to execute the remaining sentence generation process using the generated prompt; An information processing method including:

7. a first execution procedure for executing a predetermined number of sentence generation processes among a plurality of sentence generation processes executed successively using a first generation AI using a second generation AI having a higher sentence generation accuracy than the first generation AI; a generation procedure for generating a prompt including the execution result of the second generation AI by the first execution procedure, the prompt causing the first generation AI to execute the remaining sentence generation processes other than the predetermined number of sentence generation processes executed using the second generation AI; a second execution procedure for causing the first generation AI to execute the remaining sentence generation process using the generated prompt; An information processing program that causes a computer to execute the above.

Images