Risk identifying apparatus, risk identifying system, risk identifying method, and program

Abstract

To identify a risk from a target object for supporting risk assessment.SOLUTION: A target object information acquisition unit 111 acquires a target object which is a character string or an image of a target object for which a risk is identified. An information processing unit 112 generates a prompt to identify a risk estimated from the target object acquired by the target object information acquisition unit 111, and supplies the generated prompt to a generative AI to generate identification result information indicating a result of identifying a risk.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present disclosure relates to a risk identification device, a risk identification system, a risk identification method, and a program. [Background technology]

[0002] In order to prevent quality defects, accidents, etc. in product design and manufacturing, it is important to identify anticipated risks in advance and implement measures to prepare for them. Currently, techniques for assessing risk are known that utilize analysis methods such as FMEA (Failure Mode and Effects Analysis) and DRBFM (Design Review Based on Failure Mode).

[0003] For example, Patent Document 1 describes a technology that supports the creation of a new FMEA sheet by using an existing FMEA sheet. In the technology described in Patent Document 1, terms to be registered in the new FMEA sheet are selected from candidate terms registered in the existing FMEA sheet. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2018-45548 Summary of the Invention [Problem to be solved by the invention]

[0005] However, the technology described in Patent Document 1 is merely a technology for supporting term selection, and is not a technology for identifying risks. In other words, the technology described in Patent Document 1 is not a technology for identifying risks estimated from risk identification targets, which are targets for identifying risks such as character strings, images, etc. For this reason, there is a demand for a technology for identifying risks from risk identification targets in order to support risk assessment.

[0006] The present disclosure has been made in consideration of the above-mentioned problems, and aims to provide a risk identification device, a risk identification system, a risk identification method, and a program that identify risks from risk identification targets in order to assist in risk assessment. [Means for solving the problem]

[0007] In order to achieve the above-mentioned objective, the risk identification device of the present disclosure comprises a target information acquisition means for acquiring a risk identification target, which is a character string or an image of a target for which risk is to be identified, and an information processing means for generating a prompt that instructs the user to identify a risk estimated from the risk identification target acquired by the target information acquisition means, and for generating identification result information indicating the result of risk identification by supplying the generated prompt to a generation AI. [Effects of the Invention]

[0008] In the present disclosure, a prompt instructing the generation AI to identify a risk estimated from a risk identification target is supplied to the generation AI, and identification result information indicating a risk identification result is generated. Therefore, according to the present disclosure, risks can be identified from the risk identification target to support risk assessment. [Brief explanation of the drawings]

[0009] [Figure 1] Configuration diagram of a risk identification system according to the first embodiment [Figure 2] Functional configuration diagram of a risk identification system according to the first embodiment [Figure 3] A diagram showing item description information [Figure 4] FIG. 10 is a diagram showing a first prompt generated by the risk identification device according to the first embodiment. [Figure 5] FIG. 10 is a diagram showing a second prompt generated by the risk identification device according to the first embodiment. [Figure 6] 1A and 1B are diagrams showing identification result information generated by the risk identification device according to the first embodiment, where FIG. 1A shows first identification result information and FIG. 1B shows second identification result information; [Figure 7] 1 is a flowchart showing a risk identification process executed by a risk identification device according to a first embodiment. [Figure 8] Flowchart showing the prompt generation process shown in FIG. 7 [Figure 9] Configuration diagram of a risk identification system according to the second embodiment [Figure 10] Functional configuration diagram of a risk identification system according to the second embodiment [Figure 11] A diagram showing the risk identification results table as reference information [Figure 12] FIG. 10 is a diagram showing a prompt generated by a risk identification device according to a second embodiment. [Figure 13] 10 is a flowchart showing a risk identification process executed by a risk identification device according to a second embodiment. [Figure 14] A flowchart showing the prompt generation process shown in FIG. 13. [Figure 15] FIG. 10 is a diagram showing an image that is a target of risk identification according to the third embodiment. [Figure 16] FIG. 10 is a diagram showing a prompt generated by a risk identification device according to a third embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings, in which the same or corresponding parts are designated by the same reference numerals.

[0011] (Embodiment 1) FIG. 1 is a diagram showing the configuration of a risk identification system 1000 according to the first embodiment. The risk identification system 1000 is a system that identifies a risk estimated from a risk identification target, which is a character string or an image of the target for risk identification. Possible risks include, for example, safety or quality risks that occur during the design or manufacturing of an industrial product. In this case, possible risk identification targets include, for example, character strings indicating design elements, design changes, work details, etc., and images representing a work site, product, part, etc.

[0012] Design elements are components, products, etc. to be designed. For example, design elements can be a specific type of connector, a specific type of cable, or a specific type of fan blade. Design changes are changes made to the design. For example, design changes can be changing the material of the blades of a specific type of fan from one material to another, or changing the connector inserted into a specific mounting board from one type of connector to another type of connector. Work content can be inserting a specific connector into a specific mounting board, attaching a specific cable to a specific mounting board, etc.

[0013] The risk identification system 1000 includes a risk identification device 100, an inference device 200, and an imaging device 300. The risk identification device 100, the inference device 200, and the imaging device 300 are connected to each other via a communication network 700. The communication network 700 is, for example, the Internet.

[0014] The risk identification device 100 is a device that identifies an estimated risk based on a risk identification target. The risk identification device 100 outputs identification result information that indicates the risk identification results. The identification result information is, for example, information that indicates each item related to the risk for each estimated risk. The identification result information may be information corresponding to a table in which each risk is listed on the vertical axis and each item is listed on the horizontal axis.

[0015] The risk identification device 100 uses the inference device 200 to identify risks from risk identification targets. Specifically, the risk identification device 100 sends a prompt to the inference device 200 equipped with a generation AI (Artificial Intelligence), causing the inference device 200 to generate identification result information indicating the risk identification result. The risk identification device 100 outputs the generated identification result information. For example, the risk identification device 100 displays the generated identification result information. The risk identification device 100 includes a control unit 11, a memory unit 12, a display unit 13, an operation reception unit 14, and a communication unit 15.

[0016] The control unit 11 includes a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), RTC (Real Time Clock), etc. The CPU is also called a central processing unit, central arithmetic unit, processor, microprocessor, microcomputer, DSP (Digital Signal Processor), etc., and functions as a central processing unit that executes processing and calculations related to the control of the risk identification device 100. In the control unit 11, the CPU reads out programs and data stored in the ROM and uses the RAM as a work area to perform overall control of the risk identification device 100. The RTC is, for example, an integrated circuit with a timekeeping function. The CPU can determine the current date and time from the time information read out from the RTC.

[0017] The storage unit 12 includes a nonvolatile semiconductor memory such as a flash memory, an EPROM, or an EEPROM, or a hard disk drive (HDD), a solid state drive (SSD), or the like, and serves as a so-called auxiliary storage device. The storage unit 12 stores programs and data used by the control unit 11 to execute various processes. The storage unit 12 also stores data generated or acquired by the control unit 11 as a result of executing various processes.

[0018] Display unit 13 displays various images under the control of control unit 11. For example, display unit 13 displays a screen for receiving various operations from a user. Display unit 13 includes a touch screen, a liquid crystal display, an LED (Light Emitting Diode), etc. Display unit 13 displays display information.

[0019] The operation reception unit 14 receives various operations from the user and supplies information indicating the contents of the received operations to the control unit 11. The operation reception unit 14 includes a touch screen, buttons, levers, etc. The communication unit 15 communicates with devices connected to the communication network 700 in accordance with the control of the control unit 11. The communication unit 15 includes a communication interface that complies with various communication standards for connecting to the communication network 700.

[0020] The inference device 200 is a device equipped with a generative AI and capable of inferring various phenomena. The generative AI may be configured using algorithms such as Transformer, BERT (Bidirectional Encoder Representations from Transformers), GPT (Generative Pre-Training), or a combination of multiple algorithms including these. The generative AI is capable of generating various types of content such as text, still images, moving images, and audio. The generative AI learns data patterns, data relationships, and the like, and outputs new data, new information, and the like. The generative AI of this embodiment is an AI that generates text, tables, and the like.

[0021] Generative AI learns from vast amounts of data about every field that exists in the world. For this reason, generative AI can generate appropriate answers even if asked a question about a topic it has not studied. However, unlike conventional AI, generative AI is not essentially an AI that learns and answers about a specific topic, so it may generate ambiguous answers. For this reason, in order to obtain an appropriate answer using generative AI, it is extremely important to generate appropriate prompts so that the user can obtain the appropriate answer they desire. A prompt is information that corresponds to instructions, questions, etc. given to generative AI.

[0022] In this embodiment, the answer desired by the user is an answer regarding the risk estimated from the risk identification target. Therefore, the risk identification device 100 generates an appropriate prompt and supplies it to the generation AI so that an appropriate answer regarding the risk can be obtained. For example, the risk identification device 100 generates a prompt so that an answer including the items desired by the user can be obtained. Supplying the prompt to the generation AI corresponds to transmitting the prompt to the inference device 200. The inference device 200 includes a control unit 21, a memory unit 22, and a communication unit 25.

[0023] The control unit 21 includes a CPU, ROM, RAM, RTC, etc. In the control unit 21, the CPU reads out programs and data stored in the ROM and uses the RAM as a work area to perform overall control of the inference device 200. The memory unit 22 includes an HDD, SSD, etc., and serves as a so-called auxiliary storage device. The memory unit 22 stores programs and data used by the control unit 21 to execute various processes. The memory unit 22 also stores data generated or acquired by the control unit 21 executing various processes. The communication unit 25 communicates with devices connected to the communication network 700 under the control of the control unit 21. The communication unit 25 includes a communication interface that complies with various communication standards for connecting to the communication network 700.

[0024] The imaging device 300 captures images of a work site, a product, a part, etc., and generates an image. The image generated by the imaging device 300 may be a still image or a moving image. The imaging device 300 transmits the generated image to the risk identification device 100 via the communication network 700. The image generated by the imaging device 300 is used as a target for risk identification. The imaging device 300 includes a control unit that controls the overall operation of the imaging device 300, a storage unit that stores captured moving images, a CCD (Charge Coupled Device) that converts light into an electrical signal, a communication interface for connecting to the communication network 700, etc.

[0025] Next, the functions of the risk identification system 1000 will be described with reference to FIG. 2. Below, the functions of the risk identification device 100 will be mainly described. The risk identification device 100 functionally comprises a target information acquisition unit 111, an information processing unit 112, and a specified information acquisition unit 113. The inference device 200 functionally comprises an inference unit 211. Each of these functions is realized by software, firmware, or a combination of software and firmware. The software and firmware are written as programs and stored in the ROM, storage units 12, 22, etc. Then, the CPU realizes each of these functions by executing the programs stored in the ROM, storage units 12, 22, etc.

[0026] The target information acquisition unit 111 acquires a risk identification target, which is a character string or an image of a target for which a risk is to be identified. For example, the target information acquisition unit 111 acquires a character string generated based on a user operation on the operation reception unit 14. Alternatively, the target information acquisition unit 111 may acquire a character string stored in a location specified by a user operation on the operation reception unit 14. This character string is, for example, a character string indicating a design element, a design change, a work content, etc., and is a character string representing a word or a sentence.

[0027] Alternatively, the target information acquisition unit 111 acquires an image captured by the imaging device 300 from the imaging device 300 in accordance with an instruction from the user. If the image captured by the imaging device 300 is stored in another device, the target information acquisition unit 111 may acquire the image from the other device in accordance with an instruction from the user. This image is, for example, an image captured of a work site, a product, a part, etc. The target information acquisition unit 111 is an example of a target information acquisition means.

[0028] The information processing unit 112 identifies a risk estimated from the risk identification target by supplying the generation AI with a prompt that instructs the AI ​​to identify a risk based on the risk identification target. Specifically, the information processing unit 112 generates a prompt that instructs the AI ​​to identify a risk estimated from the risk identification target acquired by the target information acquisition unit 111. Then, the information processing unit 112 generates identification result information that indicates the risk identification result by supplying the generated prompt to the generation AI.

[0029] The information processing unit 112 outputs the generated identification result information. For example, the information processing unit 112 displays the generated identification result information on the display unit 13. Alternatively, the information processing unit 112 stores the generated identification result information in the storage unit 12. Alternatively, the information processing unit 112 transmits the generated identification result information to a terminal device used by the user. The information processing unit 112 is an example of an information processing means.

[0030] The information processing unit 112 generates a prompt that instructs the generation of identification result information including at least one item. For example, items used in FMEA, DRBFM, etc. can be adopted as the items. Below, the main items related to risk will be described with reference to FIG. 3. FIG. 3 shows item explanation information indicating sentences explaining each item. For example, the item explanation information is stored in the storage unit 12 and is used when generating the prompt.

[0031] As shown in FIG. 3, risk-related items include event, impact, cause, countermeasure, occurrence, severity, detection, priority, etc. An event is an event that corresponds to a risk. An event corresponds to a failure mode in FMEA. A failure mode indicates a classification according to the type of failure state, and is, for example, disconnection, short circuit, breakage, wear, deterioration of characteristics, etc. In this embodiment, an event is an essential item, and information about the event is included in the risk identification result. An impact is the impact when an event corresponding to the risk occurs. A cause is the cause of the occurrence of an event corresponding to the risk. A countermeasure is a countermeasure against the risk.

[0032] The occurrence degree is the degree of occurrence of an event corresponding to a risk. The occurrence degree is expressed, for example, as an integer value from 1 to 10. The larger the integer value of the occurrence degree, the higher the frequency of occurrence. The severity degree is the seriousness of an event corresponding to a risk when it occurs. The severity degree is expressed, for example, as an integer value from 1 to 10. The larger the integer value of the severity degree, the more serious it is.

[0033] The detection level indicates how difficult it is to detect an event corresponding to a risk. The detection level is expressed, for example, as an integer value between 1 and 10. The larger the integer value of the detection level, the more difficult it is to detect visually or with inspection equipment. The priority level is the priority for responding to a risk. The priority level is expressed, for example, as the product of the occurrence level, severity, and detection level. The larger the integer value of the priority level, the higher the priority should be given to dealing with it. The priority level corresponds to the RPN (Risk Priority Number).

[0034] The specified information acquisition unit 113 acquires item specification information that specifies items specified by the user as items to be included in the identification result information. The method by which the specified information acquisition unit 113 acquires the item specification information can be adjusted as appropriate. For example, the specified information acquisition unit 113 may display all selectable items on the display unit 13 in a selectable manner, and acquire item specification information indicating an item selected by the user operating the operation acceptance unit 14. The specified information acquisition unit 113 is an example of a specified information acquisition means.

[0035] The information processing unit 112 generates a prompt instructing the generation of identification result information including the items specified by the item designation information acquired by the designation information acquisition unit 113. Note that if the designation information acquisition unit 113 has not acquired item designation information, the information processing unit 112 generates a prompt instructing the generation of identification result information including predetermined items. Also, in this embodiment, the event corresponding to the risk is a required item. Also, in this embodiment, the inclusion of information on the content of an item in the identification result information is appropriately referred to as "the identification result information includes an item."

[0036] When it is determined that there is a possibility that information for generating identification result information is insufficient for a risk identified target, the information processing unit 112 generates a prompt to instruct the generation AI to ask again. The prompt to instruct the generation AI to ask again is a prompt to instruct the generation AI to request input of a new risk identified target when it is determined that there is a possibility that information for generating identification result information is insufficient for a risk identified target.

[0037] The method by which the information processing unit 112 determines whether there is a possibility that information is insufficient for a risk identification target can be adjusted as appropriate. For example, assume a case where it is desired to identify a risk when inserting a connector onto a mounting board. Here, if the risk identification target is "inserting a connector," it may be determined that there is a possibility that information is insufficient because the destination of the connector is unclear. Also, if the risk identification target is "inserting a connector onto a mounting board," it may be determined that there is a possibility that information is insufficient because the types of mounting board and connector are unclear.

[0038] It is considered that the information processing unit 112 and the generation AI have different criteria for determining whether or not there is insufficient information for a risk identification target. For example, even if the information processing unit 112 determines that there is insufficient information for a risk identification target, there is a possibility that the generation AI will determine that there is no insufficient information for the risk identification target. Therefore, if there is a possibility that there is insufficient information, the information processing unit 112 generates a prompt to instruct the user to ask again.

[0039] When the generation AI requests the input of a new identified risk target, the target information acquisition unit 111 acquires the new identified risk target. For example, the target information acquisition unit 111 displays on the display unit 13 a message requesting the input of a new identified risk target because an appropriate answer cannot be obtained with the current identified risk target, and acquires the new identified risk target from the user via the operation acceptance unit 14. This configuration prevents ambiguous identification result information from being generated.

[0040] Furthermore, when the information processing unit 112 determines that there is a possibility that information required to generate identification result information for a risk identification target is insufficient, it generates a prompt including examples of identification result information when information is insufficient and examples of identification result information when information is not insufficient. That is, the information processing unit 112 generates a prompt including examples of identification result information generated from a risk identification target with insufficient information and examples of identification result information generated from a risk identification target with no information insufficiency. With this configuration, the generation AI can refer to these examples of identification result information to accurately determine whether or not there is insufficient information for a risk identification target.

[0041] The inference unit 211 infers various phenomena using a trained model 221 stored in the memory unit 22. The trained model 221 is a model generated by learning from a vast amount of data on all sorts of fields that exist in the world. The inference unit 211 generates various contents according to the supplied prompts.

[0042] In this embodiment, the inference unit 211 identifies risks from risk identification targets in accordance with prompts supplied from the information processing unit 112. Specifically, the inference unit 211 identifies risks estimated from risk identification targets using the trained model 221. Then, the inference unit 211 transmits identification result information including items specified for each risk to the risk identification device 100. The inference unit 211 is an example of an inference means.

[0043] Next, specific examples of prompts generated by information processing unit 112 will be described with reference to Figures 4 and 5. Figure 4 shows a specific example of a first prompt, which is a prompt generated when the risk identification target is a character string. Figure 5 shows a specific example of a second prompt, which is a prompt generated when the risk identification target is an image.

[0044] In the example shown in FIG. 4, the first prompt includes a basic sentence, a sentence specifying a character string, a sentence specifying an item to be output, a sentence explaining each item, a sentence requesting re-input of the risk identification target, and a sentence indicating the judgment criteria for requesting re-input. The basic sentence is a sentence that indicates basic instructions to the generation AI. For example, the basic sentence is a sentence that instructs the AI ​​to identify and output a risk estimated from the specified character string. The basic sentence is a sentence written in the area indicated by dashed line 501. The sentence specifying a character string is a sentence that specifies the character string of the risk identification target. The sentence specifying a character string is a sentence written in the area indicated by dashed line 502.

[0045] The sentences specifying the items to be output are sentences specifying the items specified by the user. The sentences specifying the items to be output are sentences written in the area indicated by dashed line 503. The sentences explaining each item are sentences that specifically explain each specified item. The sentences explaining each item are generated, for example, based on item explanation information stored in memory unit 12. The sentences explaining each item are sentences written in the area indicated by dashed line 504.

[0046] The sentence requesting re-input of the risk identified target is a sentence that requests input of a risk identified target that includes more information, rather than generating ambiguous identification result information, when there is insufficient information for the risk identified target and identification result information cannot be generated. The sentence requesting re-input of the risk identified target is a sentence written in the area indicated by dashed line 505. The sentence indicating the judgment criteria for requesting re-input is a sentence indicating the specific judgment criteria for requesting re-input of the risk identified target. The sentence indicating the judgment criteria for requesting re-input is a sentence written in the area indicated by dashed line 506.

[0047] In the example shown in FIG. 5, the second prompt includes a basic sentence, a link destination of the image, a sentence specifying the items to be output, and a sentence explaining each item. The basic sentence is a sentence that indicates basic instructions to the generation AI. For example, the basic sentence is a sentence that instructs the AI ​​to identify and output the risks estimated from the specified image. The basic sentence is the sentence written in the area indicated by the dashed line 511.

[0048] The link destination of the image is the link destination of the image of the risk identification target. The link destination is basically expressed by a character string. The link destination of the image is the character string written in the area indicated by dashed line 512. In this embodiment, it is assumed that the information processing unit 112 uploads the image specified by the user to the site of this link destination before creating the prompt. Note that the information processing unit 112 may include the image in the prompt instead of including the link destination of the image in the prompt. The text specifying the items to be output is the text written in the area indicated by dashed line 513. Note that, as shown in Figures 4 and 5, the order in which the items to be output are written differs between the first prompt and the second prompt.

[0049] When the inference device 200 receives a prompt from the risk identification device 100, it analyzes the identified risk according to the sentence included in the prompt and generates identification result information. For example, the inference device 200 identifies a risk estimated from a character string specified by the prompt, and for each identified risk, it identifies the contents of the items specified by the prompt. At this time, the inference device 200 refers to the explanation of each item specified by the prompt. By including an explanation of each item in the prompt, it is expected that appropriate identification result information desired by the user will be generated.

[0050] The inference device 200 transmits the generated identification result information to the risk identification device 100. The format in which the inference device 200 outputs the identification result information can be adjusted as appropriate. The inference device 200 may output the identification result information in a matrix table format, or may output the identification result information in CSV (Comma Separated Values) format.

[0051] If the specified identified risk target is too abstract, the inference device 200 does not generate vague identification result information but requests the input of an identified risk target that includes more information. For example, if the number of risks estimated from the specified identified risk target is too large, the inference device 200 determines that the specified identified risk target is too abstract and requests the input of a more specific identified risk target from the risk identification device 100.

[0052] Referring to Figure 6, a specific example of identification result information generated by the risk identification device 100 using the inference device 200 will be described. Figure 6(A) shows a specific example of first identification result information, which is identification result information generated based on a first prompt. Figure 6(B) shows a specific example of second identification result information, which is identification result information generated based on a second prompt.

[0053] As shown in Figure 6(A), the first identification result information is information corresponding to a table showing, for each risk estimated from the character string, the event, impact, cause, countermeasure, occurrence rate, severity, detection rate, and priority. As shown in Figure 6(B), the second identification result information is information corresponding to a table showing, for each risk estimated from the image, the event, impact, severity, cause, occurrence rate, countermeasure, detection rate, and priority. The reason why the order of items listed in the first identification result information and the second identification result information is different is because the order of items specified in the first prompt and the second prompt is different.

[0054] Next, a risk identification process executed by the risk identification device 100 will be described with reference to the flowchart of Fig. 7. The risk identification process is executed, for example, in accordance with an instruction to start the risk identification process from a user.

[0055] First, the control unit 11 included in the risk identification device 100 acquires a risk identification target (step S101). For example, the control unit 11 acquires a character string specified by the user via the operation reception unit 14. Alternatively, the control unit 11 acquires an image specified by the user from the imaging device 300 or another device connected to the communication network 700. Upon completing the processing of step S101, the control unit 11 acquires item specification information (step S102). Upon completing the processing of step S102, the control unit 11 executes a prompt generation process (step S103).

[0056] The prompt generation process will be described in detail with reference to FIG. 8. First, the control unit 11 adds a basic sentence to the prompt (step S201). After completing the process of step S201, the control unit 11 adds an identified risk target to the prompt (step S202). For example, the control unit 11 adds a character string of the identified risk target to the prompt. Alternatively, the control unit 11 adds a link destination of an image of the identified risk target to the prompt. After completing the process of step S202, the control unit 11 adds a sentence indicating the items to be output to the prompt (step S203). After completing the process of step S203, the control unit 11 adds a sentence explaining each item to the prompt (step S204).

[0057] When the control unit 11 completes the processing of step S204, it determines whether or not there is a possibility that the risk identification target has insufficient information (step S205). When the control unit 11 determines that there is a possibility that the risk identification target has insufficient information (step S205: YES), it adds a sentence to the prompt requesting additional input (step S206). When the control unit 11 completes the processing of step S206, it adds a specific example of insufficient information to the prompt (step S207).

[0058] When the control unit 11 determines that there is no possibility that the risk identification target has insufficient information (step S205: NO), or when the processing of step S207 is completed, the control unit 11 completes the prompt generation processing. When the control unit 11 completes the prompt generation processing of step S103, it transmits the generated prompt to the inference device 200 (step S104). The inference device 200 analyzes the risk identification target in accordance with this prompt and generates identification result information indicating the risk identification result.

[0059] Upon completing the processing of step S104, the control unit 11 acquires identification result information from the inference device 200 (step S105). Upon completing the processing of step S105, the control unit 11 displays the identification result information (step S106). For example, the control unit 11 causes the display unit 13 to display the identification result information shown in FIG. 6. Upon completing the processing of step S106, the control unit 11 determines whether the identification result has been approved by the user (step S107).

[0060] Note that the user approves the identification result if the user obtains the desired identification result. On the other hand, the user does not approve the identification result if the user does not obtain the desired identification result. Examples of when the user does not obtain the desired identification result include when there are too many identified risks, too few identified risks, there are insufficient items, there are extra items, the contents of the items are unclear, etc. Note that the user can notify the risk identification device 100 whether or not to approve the identification result by, for example, operating the operation receiving unit 14.

[0061] If the control unit 11 determines that the identification result has not been approved by the user (step S107: NO), it returns the process to step S101. In this case, the control unit 11 accepts changes to the risk identification target, changes to the items, etc., and generates a new prompt. Then, the control unit 11 obtains new identification result information by supplying the new prompt to the inference device 200, and displays the new identification result information. If the control unit 11 determines that the identification result has been approved by the user (step S107: YES), it completes the risk identification process.

[0062] In this embodiment, a prompt instructing the generation AI to identify risks estimated from risk identification targets is supplied to the AI, and identification result information indicating the results of risk identification is generated. That is, according to this embodiment, estimated risks, causes, countermeasures, priorities, etc. are automatically identified from character strings or images indicating design elements, design changes, work details, etc. specified by the user. Therefore, according to this embodiment, risks can be identified from risk identification targets to support risk assessment.

[0063] This allows users to obtain risk information without much effort. Furthermore, users can obtain risk information even if they do not have sufficient knowledge or experience. Furthermore, because generative AI is used, it is believed that appropriate results can be obtained even if there is no accumulation of past risk-related cases.

[0064] In this embodiment, a prompt is generated to instruct the generation of identification result information including, in addition to the event, at least one of the following items for each risk: impact, cause, countermeasure, occurrence degree, severity, detection degree, and priority. According to this embodiment, identification result information suitable for risk evaluation can be obtained.

[0065] In addition, in this embodiment, a prompt is generated to instruct the generation of specific result information including the items specified by the user in the item specification information, so that the items desired by the user can be included in the specific result information.

[0066] Furthermore, in this embodiment, if the risk identification device 100 determines that there is a possibility that there is insufficient information for a risk identification target, and if the generation AI determines that there is insufficient information for the risk identification target, a prompt is generated to instruct the generation AI to request input of a new risk identification target. Therefore, this embodiment prevents inappropriate identification result information from being generated from a risk identification target with insufficient information.

[0067] Furthermore, in this embodiment, when the risk identification device 100 determines that there is a possibility that information is insufficient for a risk identification target, a prompt is generated that includes examples of identification result information generated from a risk identification target with insufficient information and examples of identification result information generated from a risk identification target with no information shortage. Therefore, according to this embodiment, the generation AI appropriately determines whether or not information is insufficient, and the generation of inappropriate identification result information and unnecessary requests to re-input the risk identification target are suppressed.

[0068] (Embodiment 2) In the first embodiment, an example in which reference information including information on similar cases is not used has been described. In the present embodiment, an example in which reference information including information on similar cases is used will be described. Note that the description of the same configurations and functions as those in the first embodiment will be omitted or simplified as appropriate.

[0069] 9, a risk identification system 1002 according to this embodiment includes a risk identification device 100, an inference device 200, an imaging device 300, and an inference device 400. The risk identification device 100, the inference device 200, the imaging device 300, and the inference device 400 are connected to each other via a communication network 700.

[0070] The inference device 400 is equipped with a dedicated AI specialized for a specific category, and is a device that infers events related to the specific category. The dedicated AI has learned deeply about the specific category and can make highly accurate inferences about events related to the specific category. In this embodiment, the dedicated AI is an AI specialized in identifying risks, and is an AI that can accurately identify risks estimated from risk identification targets. The inference device 400 includes a control unit 41, a memory unit 42, and a communication unit 45.

[0071] The control unit 41 includes a CPU, ROM, RAM, RTC, etc. In the control unit 41, the CPU reads out programs and data stored in the ROM and uses the RAM as a work area to perform overall control of the inference device 400. The memory unit 42 includes an HDD, SSD, etc., and serves as a so-called auxiliary storage device. The memory unit 42 stores programs and data used by the control unit 41 to execute various processes. The memory unit 42 also stores data generated or acquired by the control unit 41 executing various processes. The communication unit 45 communicates with devices connected to the communication network 700 under the control of the control unit 41. The communication unit 45 includes a communication interface that complies with various communication standards for connecting to the communication network 700.

[0072] Next, the functions of the risk identification system 1002 will be described with reference to FIG. 10. Below, the functions of the risk identification device 100 will be mainly described. The risk identification device 100 functionally comprises a target information acquisition unit 111, an information processing unit 112, and a specified information acquisition unit 113. The inference device 200 functionally comprises an inference unit 211. The inference device 400 functionally comprises an inference unit 411. Each of these functions is realized by software, firmware, or a combination of software and firmware. The software and firmware are written as programs and stored in the ROM, storage units 12, 22, 42, etc. The CPU then executes the programs stored in the ROM, storage units 12, 22, 42, etc. to realize each of these functions.

[0073] The inference unit 411 infers risk-related events using a trained model 421 stored in the storage unit 42. The trained model 421 is a model that has learned about risk. For example, the trained model 421 is a model that has learned the correspondence between identified risk targets and identification result information using a huge amount of training data including the identified risk targets and identification result information corresponding to the identified risk targets. The inference unit 411 can use the trained model 421 to obtain the identification result information corresponding to the identified risk targets from the supplied identified risk targets.

[0074] The database 422 is a database constructed in the storage unit 42. The database 422 is a database in which information obtained using the inference unit 411 is accumulated. Specifically, the database 422 is a database in which information indicating the correspondence between identified risk targets and identification result information is accumulated. The identified risk targets accumulated in the database 422 are identified risk targets accumulated in the past from the time of use of the risk identification device 100, and therefore will be referred to as past identified risk targets hereinafter as appropriate. The identified result information accumulated in the database 422 is identified risk result information accumulated in the past from the time of use of the risk identification device 100, and therefore will be referred to as past identified risk result information hereinafter as appropriate.

[0075] In this embodiment, the database 422 stores a large number of risk identification result tables, each including past risk identification targets and past identification result information. In this embodiment, the risk identification result table corresponds to an FMEA table. FIG. 11 shows table 800, which is an example of a risk identification result table. In FIG. 11, the information in the area surrounded by dashed line 801 corresponds to past risk identification targets, and the information in the area surrounded by dashed line 802 corresponds to past identification result information. In this embodiment, it is assumed that a large number of risk identification result tables have been stored in the database 422 before the risk identification device 100 executes the risk identification process.

[0076] In this embodiment, the information processing unit 112 refers to a database 422 that stores past risk identification targets, which are character strings or images in which risks have been identified in the past, in association with past identification result information that indicates the results of risk identification estimated from the past risk identification targets. The information processing unit 112 identifies similar identified targets, which are past risk identification targets similar to the risk identification target acquired by the target information acquisition unit 111, from the past risk identification targets stored in the database 422. Note that the similar identified targets may be the same as the risk identification target acquired by the target information acquisition unit 111.

[0077] The method for determining the similarity between an identified risk object and a past identified risk object can be adjusted as appropriate. For example, if the identified risk object and the past identified risk object contain many common or similar terms, the identified risk object and the past identified risk object may be determined to be similar. For example, 3 points may be assigned for each common term and 1 point for each set of similar terms, and the higher the normalized total score, the higher the similarity is considered to be. As a normalization method, for example, a method of dividing by the total number of terms included in the identified risk object and the past identified risk object may be considered.

[0078] For example, in the expressions "attaching a cable to a mounting board" and "inserting a connector onto a mounting board," the term "mounting board" is common, the terms "cable" and "connector" are similar, and the terms "attach" and "insertion" are similar. In this case, the total score is 5 points, and the total number of terms is 6, so the normalized score is 5 / 6 points. In this case, "attaching a cable to a mounting board" and "inserting a connector onto a mounting board" may be determined to be similar. Note that the information processing unit 112 can determine whether two terms match and whether the two terms are similar by, for example, referring to similarity determination information. The similarity determination information is information indicating whether the two terms match and whether the two terms are similar for each combination of two terms, and is stored, for example, in the storage unit 12.

[0079] Furthermore, the information processing unit 112 identifies similar identification result information, which is past identification result information associated with the similar identified target, from the past identification result information stored in the database 422. In this way, the information processing unit 112 identifies similar identified targets, which are past risk identified targets similar to the risk identified target acquired by the target information acquisition unit 111, and similar identification result information, which is past identification result information associated with the similar identified target.

[0080] The information processing unit 112 then generates a prompt instructing the generation of identification result information by referring to reference information including the similar identified target and the similar identified result information. Meanwhile, the generation AI generates identification result information from the supplied risk identified target by referring to the reference information. For example, the generation AI can refer to the items, output format, terminology, etc. indicated in the reference information. Furthermore, when the risk identified target and the similar identified target match, the generation AI can generate identification result information corresponding to the risk identified target by supplementing the items missing in the similar identified result information.

[0081] Note that the items, output format, terminology, etc. required for the identification result information often differ depending on the company, department, etc. to which the user who uses the identification result information belongs. On the other hand, if the inference device 400 is operated by the company, department, etc. to which the user of the risk identification device 100 belongs, the items, output format, terminology, etc. of the past identification result information stored in the database 422 are likely to be similar to the items, output format, terminology, etc. of the identification result information desired by the user. For this reason, it is considered that the generation AI can generate the identification result information desired by the user by referring to reference information including the past identification result information accumulated in the database 422.

[0082] In this embodiment, the risk identification target acquired by the target information acquisition unit 111 is "insertion of a connector onto a mounting board." The similar identification target is "attachment of a cable to a mounting board" shown within dashed line 801 in Fig. 11. The similar identification result information is information corresponding to the table shown within dashed line 802 in Fig. 11. The reference information is information corresponding to the risk identification result table shown in Fig. 11.

[0083] Next, a specific example of a prompt generated by the information processing unit 112 will be described with reference to Fig. 12. In the example shown in Fig. 12, the prompt includes a basic sentence, a sentence specifying a character string, a sentence specifying an item to be output, a sentence explaining each item, a basic sentence related to reference information, a link destination of the reference information, and additional sentences related to the reference information.

[0084] The basic sentence regarding reference information is a sentence that provides basic instructions based on the reference information to the generation AI. For example, the basic sentence regarding reference information is a sentence that instructs the generation AI to analyze risks by referring to the reference information. The basic sentence regarding reference information is a sentence written in the area indicated by the dashed line 521.

[0085] The link destination of the reference information is a link destination of reference information to be used as a reference when analyzing risks. The link destination of the reference information is a character string written in the area indicated by dashed line 522. In this embodiment, it is assumed that information processing unit 112 uploads the reference information to this linked site before creating the prompt. Note that information processing unit 112 may include the reference information in the prompt instead of including the link destination of the reference information in the prompt.

[0086] The additional text regarding the reference information is a text that indicates additional instructions based on the reference information to the generation AI. For example, the additional text regarding the reference information is a text that instructs the generation AI to generate identification result information with the specified item added if an item specified by the user is not in the risk identification result table. The additional text regarding the reference information is a text written in the area indicated by the dashed line 523.

[0087] The identification result information may or may not include an identified risk target. For example, the identification result information may be information corresponding to a table in which the information within dashed lines 801 and 802 in the risk identification result table shown in Figure 11 has been replaced, that is, information in which an identified risk target has been added to the information shown in Figure 6(A). Alternatively, the identification result information may be information corresponding to a table in which the information within dashed line 801 in the risk identification result table shown in Figure 11 has been deleted and the information within dashed line 802 has been replaced, that is, information shown in Figure 6(A).

[0088] Next, the risk identification process executed by the risk identification device 100 according to this embodiment will be described with reference to the flowchart of FIG.

[0089] First, the control unit 11 acquires a risk identification target (step S101). After completing the processing of step S101, the control unit 11 acquires item specification information (step S102). After completing the processing of step S102, the control unit 11 searches for similar cases (step S102A). For example, the control unit 11 determines whether or not there is a risk identification result table stored in the database 422 that includes "insertion of a connector onto a mounting board" as a past risk identification target.

[0090] If the control unit 11 determines that such a risk identification result table exists, it identifies this risk identification result table as a similar case. The control unit 11 identifies "attaching a cable to a mounting board" included in the identified risk identification result table as a similar identification target. The control unit 11 also identifies the past identification result information included in the identified risk identification result table as similar identification result information. When the control unit 11 completes the processing of step S102A, it executes a prompt generation process (step S103).

[0091] The prompt generation process according to this embodiment will be described in detail with reference to Figure 14. The processes from step S201 to step S204 are the same as those described in the first embodiment. After completing the process of step S204, the control unit 11 determines whether there are any similar cases (step S204A). That is, the control unit 11 determines whether a risk identification result table containing "insertion of connector onto mounting board" has been found in the search for similar cases.

[0092] When the control unit 11 determines that there is a similar case (step S204A: YES), it adds a basic sentence regarding reference information to the prompt (step S211). That is, the control unit 11 adds a sentence to the prompt instructing that information corresponding to the risk identification result table of the similar case be used as reference information. When the control unit 11 completes the processing of step S211, it adds the reference information to the prompt (step S212). For example, the control unit 11 adds a link to the risk identification result table of the similar case to the prompt.

[0093] When the control unit 11 completes the processing of step S212, it adds additional text related to reference information to the prompt (step S213). When the control unit 11 determines that there are no similar cases (step S204A: NO), or when it completes the processing of step S213, it completes the prompt generation processing. When the control unit 11 completes the prompt generation processing of step S103, it transmits the generated prompt to the inference device 200 (step S104). Note that the processing from step S104 onwards is as described in embodiment 1.

[0094] In this embodiment, a prompt is generated to instruct the generation of identification result information by referring to reference information including similar identified targets and similar identification result information. In other words, identification result information for the current risk identified target is generated based on past identification result information corresponding to past risk identified targets similar to the current risk identified target. Therefore, according to this embodiment, appropriate identification result information can be generated.

[0095] (Embodiment 3) In the first embodiment, an example was described in which, when the target for risk identification is an image, a prompt is generated to instruct the user to identify the risk from the entire image. In the present embodiment, an example is described in which, when the target for risk identification is an image, a prompt is generated to instruct the user to identify the risk from a portion of the entire image. Note that the description of the same configurations and functions as those in the first and second embodiments will be omitted or simplified as appropriate.

[0096] In this embodiment, the target information acquisition unit 111 acquires an entire image including a designated image that designates a partial image, which is a specific image portion, as a risk identification target. The information processing unit 112 generates a prompt that instructs the user to identify a risk from the partial image designated by the designated image of the entire image. This will be specifically described below with reference to FIG. 15.

[0097] FIG. 15 shows image 900, which is an overall image acquired by the target information acquisition unit 111. Image 900 is a still image showing two workers performing maintenance on a mold. Image 900 includes image 910, which is a designated image for designating a partial image. The partial image is an image of the portion of image 900 that is surrounded by image 910. The partial image is an image showing the worker in the foreground performing maintenance on one of the four openings in the mold. The information processing unit 112 generates a prompt that instructs the user to identify a risk from the partial image surrounded by image 910.

[0098] With this configuration, the user can easily understand the risk estimated from a partial image of the entire image. In other words, when the user cannot prepare an image showing only the partial image, the user can specify the partial image in the entire image by adding a designated image to the entire image. One method of adding a designated image to the entire image is to use image editing software to add a designated image as a marker that surrounds the partial image in the entire image. The designated image may not be an image that surrounds the partial image, but may be an image of an arrow that indicates the partial image. Note that when a designated image is not added to the entire image, not only the risk related to the partial image but also the risk related to the image of other parts other than the partial image is identified.

[0099] A specific example of a prompt generated by information processing unit 112 according to this embodiment will be described with reference to Fig. 16. In the example shown in Fig. 16, the prompt includes a basic sentence, a link destination for the image, a sentence specifying a partial image, a sentence specifying items to be output, and a sentence explaining each item. The sentence specifying the partial image is a sentence specifying that a risk should be identified from a specified partial image of the entire image. The sentence specifying the partial image is a sentence written in the area indicated by dashed line 531.

[0100] In this embodiment, a prompt is generated to instruct the user to identify a risk from a partial image designated by a designated image of the entire image. Thus, according to this embodiment, the entire image can be used to identify a risk estimated from the partial image.

[0101] (Variation) Although the embodiments have been described above, modifications and applications in various forms are possible. It is up to the discretion of the individual to adopt any of the configurations, functions, and operations described in the above embodiments. Furthermore, in addition to the above-described configurations, functions, and operations, additional configurations, functions, and operations may be adopted. Furthermore, the configurations, functions, and operations described in the above embodiments can be freely combined.

[0102] In the first embodiment, an example has been described in which the identification result information is basically information corresponding to an FMEA table. The identification result information may also be information corresponding to a DRBFM table. Note that the user can specify FMEA or DRBFM via the operation reception unit 14 using a method similar to the method for specifying an item. Note that FMEA is a systematic analysis method for potential failures aimed at preventing breakdowns, malfunctions, etc. DRBFM is a tool developed based on the philosophy that making changes to an existing engineering design that has already proven successful will cause design problems.

[0103] In the first embodiment, an example has been described in which the inference device 200 equipped with the generation AI is an internal component of the risk identification system 1000, but the inference device 200 may be an external component of the risk identification system 1000. Even in this case, the risk identification device 100 can cause the inference device 200, which is an external component, to output identification result information by supplying a prompt to the inference device 200. With this configuration, it is possible to obtain identification result information by utilizing various artificial intelligence services external to the risk identification system 1000.

[0104] In the above-described embodiments, the control units 11, 21, and 41 function as the units shown in FIGS. 2 and 10 by the CPU executing programs stored in the ROM or the storage units 12, 22, and 42. However, in the present disclosure, the control units 11, 21, and 41 may be dedicated hardware. Dedicated hardware may be, for example, a single circuit, a composite circuit, a programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a combination thereof. When the control units 11, 21, and 41 are dedicated hardware, the functions of each unit may be realized by individual hardware, or the functions of each unit may be realized collectively by a single piece of hardware. Furthermore, some of the functions of each unit may be realized by dedicated hardware, and the other functions may be realized by software or firmware. In this manner, the control units 11, 21, and 41 may realize the above-described functions by hardware, software, firmware, or a combination thereof.

[0105] By applying an operating program that defines the operation of the risk identification device according to the present disclosure to a computer such as an existing personal computer or information terminal device, it is possible to cause the computer to function as the risk identification device according to the present disclosure. Furthermore, such a program can be distributed in any manner, and may be distributed by being stored on a computer-readable recording medium such as a CD-ROM (Compact Disk ROM), a DVD (Digital Versatile Disk), an MO (Magneto Optical Disk), or a memory card, or may be distributed via a communication network such as the Internet.

[0106] The present disclosure allows various embodiments and modifications without departing from the broad spirit and scope of the present disclosure. Furthermore, the above-described embodiments are intended to explain the present disclosure and do not limit the scope of the present disclosure. That is, the scope of the present disclosure is defined by the claims, not the embodiments. Various modifications made within the scope of the claims and the meaning of equivalent disclosures are considered to be within the scope of the present disclosure.

[0107] Various aspects of the present disclosure are summarized below as appendices.

[0108] (Appendix 1) a target information acquisition means for acquiring a risk identification target, which is a character string or an image of a target for identifying a risk; and an information processing means for generating a prompt instructing the user to identify a risk estimated from the risk identification target acquired by the target information acquisition means, and for generating identification result information indicating the result of the risk identification by supplying the generated prompt to a generation AI. Risk identification device. (Appendix 2) the information processing means generates the prompt instructing to generate the identification result information including, for each risk estimated from the risk identification target, in addition to an event corresponding to the risk, at least one item of an impact when the event occurs, a cause of the event, a measure against the risk, a degree of occurrence of the event, a severity when the event occurs, a detection degree indicating the difficulty of detecting the event, and a priority of a response to the risk. The risk identification device described in Appendix 1. (Appendix 3) a specified information acquisition means for acquiring item specification information for specifying an item specified by a user as an item to be included in the identification result information; the information processing means generates the prompt instructing to generate the identification result information including the item specified by the item specification information acquired by the specification information acquisition means. 1. A risk identification device according to claim 1 or 2. (Appendix 4) When the information processing means determines that there is a possibility that information for generating the identification result information is insufficient for the risk identified target, if it determines that information for generating the identification result information is insufficient for the risk identified target, it generates the prompt that instructs the generation AI to request input of a new risk identified target, The target information acquisition means acquires the new risk identified target when the generation AI requests input of the new risk identified target. 10. The risk identification device of claim 1, 2, 3, or 4. (Appendix 5) When the information processing means determines that there is a possibility that the risk identified object lacks information for generating the identification result information, it generates the prompt including examples of identification result information to be generated from the risk identified object lacking information and examples of identification result information to be generated from the risk identified object not lacking information. The risk identification device described in Appendix 4. (Appendix 6) the information processing means refers to a database in which past risk identification targets, which are character strings or images in which risks have been identified in the past, and past identification result information, which indicates identification results of risks estimated from the past risk identification targets, are stored in association with each other; the information processing means identifies similar identified targets, which are past risk identification targets similar to the risk identification target acquired by the target information acquisition means, and similar identified result information, which is past identification result information associated with the similar identified targets; and generates the prompt instructing the user to generate the identification result information by referring to reference information including the similar identified targets and the similar identified result information. 6. A risk identification device according to any one of appendices 1 to 5. (Appendix 7) The target information acquisition means acquires an entire image including a designated image that designates a partial image that is a specific image portion as the risk identification target, the information processing means generates the prompt instructing the user to identify the risk from the partial image of the entire image designated by the partial designation image. 10. The risk identification device of any one of appendices 1 to 6. (Appendix 8) A risk identification system comprising an inference device equipped with a generation AI and a risk identification device communicating with the inference device, The risk identification device a target information acquisition means for acquiring a risk identification target, which is a character string or an image of a target for identifying a risk; an information processing means for generating a prompt instructing the inference device to identify a risk estimated from the risk identification target acquired by the target information acquisition means, and transmitting the generated prompt to the inference device; the inference device includes inference means for identifying the risk from the risk identification target in accordance with the prompt and transmitting identification result information indicating the risk identification result to the risk identification device; the information processing means outputs the identification result information received from the inference device. Risk identification system. (Appendix 9) Acquire a risk identification target, which is a character string or an image of the target for identifying a risk; generating a prompt that instructs the user to identify a risk estimated from the acquired risk identification target, and generating identification result information that indicates the result of identifying the risk by supplying the generated prompt to a generation AI; Risk identification methods. (Appendix 10) Computer, a target information acquisition means for acquiring a risk identification target, which is a character string or an image of a target for identifying a risk; generating a prompt instructing the target information acquisition means to identify a risk estimated from the risk identification target acquired by the target information acquisition means, and supplying the generated prompt to a generation AI, thereby functioning as an information processing means for generating identification result information indicating the result of risk identification; program. [Industrial Applicability]

[0109] The present disclosure is applicable to risk identification systems. [Explanation of symbols]

[0110] 11,21,41 Control unit, 12,22,42 Memory unit, 13 Display unit, 14 Operation reception unit, 15,25,45 Communication unit, 100 Risk identification device, 111 Target information acquisition unit, 112 Information processing unit, 113 Designated information acquisition unit, 200,400 Inference device, 211,411 Inference unit, 221,421 Trained model, 422 Database, 300 Imaging device, 501,502,503,504,511,512,513,521,522,523,531,801,802 Dashed line, 700 Communication network, 800 Table, 900,910 Image, 1000,1002 Risk identification system

Claims

1. a target information acquisition means for acquiring a risk identification target, which is a character string or an image of a target for identifying a risk; and an information processing means for generating a prompt instructing the user to identify a risk estimated from the risk identification target acquired by the target information acquisition means, and supplying the generated prompt to a generation AI to generate identification result information indicating the result of the risk identification. Risk identification device.

2. the information processing means generates the prompt instructing to generate the identification result information including, for each risk estimated from the risk identification target, in addition to an event corresponding to the risk, at least one item of an impact when the event occurs, a cause of the event, a measure against the risk, a degree of occurrence of the event, a severity when the event occurs, a detection degree indicating the difficulty of detecting the event, and a priority of a response to the risk. The risk identification device according to claim 1 .

3. a specified information acquisition means for acquiring item specification information for specifying an item specified by a user as an item to be included in the identification result information; the information processing means generates the prompt instructing to generate the identification result information including the item specified by the item specification information acquired by the specification information acquisition means. The risk identification device according to claim 1 or 2.

4. When the information processing means determines that there is a possibility that information for generating the identification result information is insufficient for the risk identified object, if it determines that information for generating the identification result information is insufficient for the risk identified object, it generates the prompt that instructs the generation AI to request input of a new risk identified object, The target information acquisition means acquires the new risk identified target when the generation AI requests input of the new risk identified target. The risk identification device according to claim 1 or 2.

5. When the information processing means determines that there is a possibility that the risk identified object lacks information for generating the identification result information, it generates the prompt including examples of identification result information to be generated from the risk identified object lacking information and examples of identification result information to be generated from the risk identified object not lacking information. The risk identification device according to claim 4.

6. the information processing means refers to a database in which past risk identification targets, which are character strings or images in which risks have been identified in the past, and past identification result information, which indicates identification results of risks estimated from the past risk identification targets, are stored in association with each other; the information processing means identifies similar identified targets, which are past risk identification targets similar to the risk identification target acquired by the target information acquisition means, and similar identified result information, which is past identification result information associated with the similar identified targets; and generates the prompt instructing the user to generate the identification result information by referring to reference information including the similar identified targets and the similar identified result information. The risk identification device according to claim 1 or 2.

7. The target information acquisition means acquires an entire image including a designated image that designates a partial image that is a specific image portion as the risk identification target, the information processing means generates the prompt instructing the user to identify the risk from the partial image designated by the designated image of the entire image. The risk identification device according to claim 1 or 2.

8. A risk identification system comprising an inference device equipped with a generating AI and a risk identification device communicating with the inference device, The risk identification device a target information acquisition means for acquiring a risk identification target, which is a character string or an image of a target for identifying a risk; an information processing means for generating a prompt instructing the inference device to identify a risk estimated from the risk identification target acquired by the target information acquisition means, and transmitting the generated prompt to the inference device; the inference device includes inference means for identifying the risk from the risk identification target in accordance with the prompt and transmitting identification result information indicating the risk identification result to the risk identification device; the information processing means outputs the identification result information received from the inference device. Risk identification system.

9. Acquire a risk identification target, which is a character string or an image of the target for identifying a risk; generating a prompt instructing to identify a risk estimated from the acquired risk identification target, and generating identification result information indicating the result of identifying the risk by supplying the generated prompt to a generation AI; Risk identification methods.

10. Computer, a target information acquisition means for acquiring a risk identification target, which is a character string or an image of a target for identifying a risk; The information processing means generates a prompt that instructs the target information acquisition means to identify a risk estimated from the risk identification target acquired by the target information acquisition means, and supplies the generated prompt to a generation AI to generate identification result information that indicates the result of the risk identification. program.

Images