Human resource development support device

Abstract

To provide a human resource development support device using generative AI which efficiently performs objective evaluations with little variation.SOLUTION: A human resource development support device 1 includes: an interview execution prompt provision module 13 which provides an interview execution prompt which instructs a first generative AI 2 to conduct an interview for human resource evaluation and then instructs the first generative AI 2 to generate a first evaluation value for a subject on the basis of responses from the subject during the interview; a model creation module which creates a model representing the relation between explanatory variables and objective variables, using second evaluation values of respective learning subjects as explanatory variables and pass / fail determination results of the respective learning subjects as objective variables; and a pass / fail determination module 15 including a score calculation module which inputs the first evaluation value of the subject into the model to calculate a score of the subject.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a human resource development support device that uses the results of interviews using generation AI to perform machine learning and calculate scores, thereby supporting appropriate human resource development. [Background technology]

[0002] In-house human resource development involves conducting interviews, providing feedback on the results, and having employees undergo training. With the spread of remote work, online interviews are also becoming common. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 7333116 Summary of the Invention [Problem to be solved by the invention]

[0004] Patent document 1 describes that questionnaire information showing the results of a question survey of the interviewee is obtained before the interview, and the interview organizer selects questions to be used during the interview from a number of candidate questions for the interviewee.

[0005] However, in interviews conducted by humans, variations in evaluation accuracy are inevitable. Even if scoring is done according to rules, human judgment is involved when applying the rules. In particular, when trying to make a comprehensive judgment based on a variety of factors, variations in judgment can occur even when the judge is the same person. Furthermore, whether the interview is face-to-face or remote, it is necessary to coordinate the schedules of both participants, which can make it difficult to proceed efficiently.

[0006] One aspect of the present invention relates to an efficient, objective evaluation with little variance, in which a generation AI interviews a subject, creates an evaluation value for the subject, and calculates the subject's score using the results of machine learning. [Means for solving the problem]

[0007] A human resource development support device according to one aspect of the present invention comprises: an interview execution prompt providing module that instructs a first generation AI to conduct an interview for talent evaluation and provides an interview execution prompt that instructs the first generation AI to create a first evaluation value of the subject based on answers from the subject in the interview; a model creation module that creates a model indicating a relationship between the explanatory variables and the objective variables, using the second evaluation values ​​of each of a plurality of learning objects as explanatory variables and the pass / fail judgment results of each of the plurality of learning objects as objective variables; a score calculation module that inputs the first evaluation value of the subject into the model to calculate a score of the subject; Includes: [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 shows a human resource development assisting device 1 according to the first embodiment and an external device connected to the human resource development assisting device 1. As shown in FIG. [Figure 2] FIG. 2 shows the functions and operations of the human resource development assisting device 1 according to the first embodiment. [Figure 3A] FIG. 3A shows an example of an evaluation item creation prompt for causing the first generation AI 2 to create an evaluation item (evaluation perspective) in the first embodiment. [Figure 3B] FIG. 3B shows the response from the first generation AI2 to the assessment item creation prompt shown in FIG. 3A. [Figure 3C] FIG. 3C shows an example of an evaluation item creation prompt for causing the first generation AI2 to create an evaluation item (check item) in the first embodiment. [Figure 3D]FIG. 3D shows the response from the first generation AI2 to the assessment item creation prompt shown in FIG. 3C. [Figure 3E] FIG. 3E shows an example of a question creation prompt for causing the first generation AI2 to create a question in the first embodiment. [Figure 3F] FIG. 3F shows the response from the first generation AI2 to the question-generating prompt shown in FIG. 3E. [Figure 4A] FIG. 4A shows an example of an interview execution prompt for causing the first generated AI2 to conduct an interview and generate a first evaluation value in the first embodiment. [Figure 4B] FIG. 4B shows an example of the progress of an interview that is carried out when the interview execution prompt, evaluation items, and question sentences shown in FIG. 4A are sent to the first generation AI2. [Figure 5A] FIG. 5A shows an example of learning data used for creating a model in the first embodiment. [Figure 5B] FIG. 5B shows an example of determination data used for score calculation in the first embodiment. [Figure 5C] FIG. 5C shows the details of the pass / fail determination module 15. [Figure 6] FIG. 6 shows the functions and operations of a human resource development assisting device 1a according to the second embodiment. [Figure 7] FIG. 7 shows an example of a training execution prompt for causing the second generated AI 2a to execute training by role-playing in the second embodiment. [Figure 8A] FIG. 8A shows Example 1 in which "prevention of power harassment" is set as the training purpose and "subordinate" is set as the trainer's role. [Figure 8B] FIG. 8B shows a scenario generated according to Example 1. [Figure 8C] FIG. 8C shows an example of a training run according to Example 1. [Figure 9A] FIG. 9A shows Example 2 in which "improvement of management skills" is set as the training purpose and "supervisor or customer" is set as the trainer's role. [Figure 9B]FIG. 9B shows a scenario generated according to Example 2. [Figure 9C] FIG. 9C shows an example of training execution according to Example 2. [Figure 10A] FIG. 10A shows Example 3 in which "improving customer satisfaction" is set as the training purpose and "customer" is set as the trainer's role. [Figure 10B] FIG. 10B shows a scenario generated according to Example 3. [Figure 10C] FIG. 10C shows an example of training execution according to Example 3. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Each embodiment described below shows an example of the present invention and does not limit the content of the present invention. Furthermore, not all of the configurations and operations described in each embodiment are necessarily essential as the configurations and operations of the present invention. Note that the same components are given the same reference numerals, and redundant explanations will be omitted.

[0010] <1. First embodiment> <1-1.Configuration> FIG. 1 shows a human resource development support device 1 according to the first embodiment and external devices connected to the human resource development support device 1. The human resource development support device 1 is a computer system including a CPU, memory, etc. (not shown). The human resource development support device 1 may be configured with a single computer or multiple computers connected via a network. The human resource development support device 1 is connected to external devices such as a first generation AI 2, a subject terminal 3, and a database 4.

[0011] The first generation AI 2 includes a large-scale language model (LLM). The large-scale language model is a language model constructed using large amounts of text data and deep learning technology, and processes tasks such as text generation, translation, and answering questions in response to prompts sent from the human resource development support device 1. The large-scale language model preferably includes an attention mechanism. The attention mechanism extracts important parts from the input and significantly contributes to improving the processing speed and accuracy of the large-scale language model.

[0012] The subject terminal 3 is a terminal operated by a subject, such as an employee, who is being interviewed, and is a computer system equipped with an input device, an output device, a CPU, a memory, etc. (not shown). The subject terminal 3 can access the human resource development support device 1, receives questions from the human resource development support device 1, outputs them to the output device, and transmits answers from the subject entered into the input device to the human resource development support device 1.

[0013] The database 4 stores learning data and other data. The database 4 is not limited to one in which data is stored in a single storage device, but may be one in which data is stored distributed across multiple storage devices. The human resource development assistance device 1 acquires various data from the database 4 and performs processes such as model creation.

[0014] The human resource development support device 1 sends a prompt to the first generation AI 2 and obtains output from the first generation AI 2. The human resource development support device 1 sends the question output from the first generation AI 2 to the subject terminal 3 and sends the answer received from the subject terminal 3 to the first generation AI 2. If the question and answer data exchanged via the human resource development support device 1 includes voice, image, or video, the first generation AI 2 includes a system required depending on the data format, such as a voice synthesis system, a voice recognition system, an image generation system, or an image recognition system.

[0015] <1-2. Functions and operations> 2 shows the functions and operations of the human resource development assistance device 1 according to the first embodiment. The human resource development assistance device 1 includes an evaluation item creation prompt providing module 11, a question creation prompt providing module 12, an interview execution prompt providing module 13, and a pass / fail determination module 15. These modules are implemented by loading programs into the memory included in the human resource development assistance device 1 and executing them by the CPU.

[0016] The evaluation item creation prompt providing module 11 provides an evaluation item creation prompt that instructs the first generation AI 2 to consider evaluation items for talent evaluation, and transmits it to the first generation AI 2. Specific examples of the evaluation item creation prompt and its response will be described later with reference to Figures 3A to 3D.

[0017] The question creation prompt providing module 12 provides a question creation prompt that instructs the first generation AI 2 to create a question to be used in an interview for talent evaluation based on the evaluation items received from the first generation AI 2, and sends the question creation prompt to the first generation AI 2. Specific examples of the question creation prompt and its response will be described later with reference to Figures 3E and 3F.

[0018] The interview execution prompt providing module 13 provides an interview execution prompt that instructs the first generation AI2 to conduct an interview for talent evaluation using the question generated by the first generation AI2 and to create a first evaluation value for the subject based on the answers from the subject in the interview, and sends the prompt to the first generation AI2. Specific examples of the interview execution prompt and its response will be described later with reference to Figures 4A and 4B. The interview is conducted by the human resource development support device 1 mediating between the questions output by the first generation AI2 and the answers entered into the subject terminal 3.

[0019] The first generation AI2 that creates evaluation items in response to evaluation item creation prompts, the first generation AI2 that creates questions in response to question creation prompts, and the first generation AI2 that conducts interviews in response to interview execution prompts may be separate generation AIs that include separate large-scale language models.

[0020] Pass / fail determination module 15 creates a model based on the learning data stored in database 4, calculates a score by inputting the first evaluation value received from first generation AI 2 into the model, and determines whether the subject passes or fails. Model creation will be described later with reference to FIG. 5A, score calculation with reference to FIG. 5B, and details of pass / fail determination module 15 with reference to FIG. 5C. The subject's pass / fail result can be viewed by a supervisor or a human resources consultant or sent to subject terminal 3, and is used for human resource development.

[0021] <1-3. Specific examples> <1-3-1. Creating evaluation items> 3A and 3C show examples of evaluation item creation prompts for causing the first generation AI2 to create evaluation items in the first embodiment. The evaluation item creation prompt can be composed of two stages: a prompt (FIG. 3A) asking for "evaluation perspectives" for personnel evaluation, and a prompt (FIG. 3C) asking for "check items" that subdivide the evaluation perspectives. The prompt asking for evaluation perspectives may be a pre-stored template, or it may be generated by adding a category or job content set by a supervisor or a human resources consultant, such as "project management," to the template. The prompt asking for check items may be generated by adding evaluation perspectives obtained from the first generation AI2 to a pre-stored template. Alternatively, the prompt may ask for both evaluation perspectives and check items simultaneously, such as "Please list evaluation perspectives and propose check items for each evaluation perspective."

[0022] Figures 3B and 3D show responses from the first generation AI2 to the evaluation item creation prompts shown in Figures 3A and 3C, respectively. The first generation AI2 proposes evaluation perspectives A to D shown in Figure 3B and check items A-1 to D-3 shown in Figure 3D for each of evaluation perspectives A to D. The evaluation items proposed by the first generation AI2 may be modified by a human being, such as a supervisor or human resources consultant, or they may be created by a human being without having the first generation AI2 create the evaluation items.

[0023] <1-3-2. Creating Questions> FIG. 3E shows an example of a question creation prompt for causing the first generation AI2 to create a question in the first embodiment. The question creation prompt requests a question sentence for evaluating the subject for each item and may be stored in advance as a template. Note that when the evaluation item creation prompt and the question creation prompt are sent to the same first generation AI2, the evaluation items created in response to the evaluation item creation prompt are assumed to be inherited within the first generation AI2. When the evaluation item creation prompt and the question creation prompt are sent to different generation AIs, the evaluation items created in response to the evaluation item creation prompt must be included in the question creation prompt.

[0024] The question creation prompt can be provided with multiple types of questions (question type 1 and type 2), and the first generation AI 2 can be made to generate questions for each item and type. Furthermore, if there is evaluation value data obtained from past interviews with the same subject, the first generation AI 2 can be instructed to adjust the difficulty of the questions taking into account the evaluation value. Furthermore, a prompt that combines the evaluation item creation prompt and the question creation prompt can be used to simultaneously create evaluation perspectives, check items, and questions.

[0025] Figure 3F shows a response from the first generation AI2 to the question creation prompt shown in Figure 3E. Because the first generation AI2 generates the questions, the questions necessary for the interview can be prepared in a short time, even if there are many items for which questions need to be created. After obtaining the questions for each evaluation item from the first generation AI2, the human resource development support device 1 stores the evaluation items and questions in the database 4, for example, in CSV (comma-separated values) format.

[0026] <1-3-3. Conducting interviews> 4A shows an example of an interview execution prompt for causing the first generation AI2 to conduct an interview and generate a first evaluation value in the first embodiment. The interview execution prompt includes a command statement instructing the first generation AI2 to conduct an interview as an interviewer and generate a first evaluation value and evaluation comments. The evaluation items and questions can be provided separately in a format such as a CSV file.

[0027] FIG. 4B shows an example of the progress of an interview that is executed when the interview execution prompt, evaluation items, and questions shown in FIG. 4A are sent to the first generation AI2. The first generation AI2 outputs questions one by one, starting with the first question A-1, according to the list of questions. The human resource development support device 1 transmits the questions output from the first generation AI2 to the subject terminal 3. The human resource development support device 1 accepts the subject's answers from the subject terminal 3 and transmits them to the first generation AI2. Upon receiving the answers from the subject, the first generation AI2 creates a first evaluation value and evaluation comment and writes them to a CSV file. The first evaluation value and evaluation comment may be transmitted to the subject terminal 3, but at least a portion of the first evaluation value may not be transmitted to the subject terminal 3 and may be used only for calculating the score. After writing the first evaluation value and evaluation comment to the CSV file, the first generation AI2 outputs the next question A-2. The interview proceeds in a similar manner, and the first evaluation value and evaluation comment are accumulated. When the interview is completed, the human resource development assisting device 1 receives the CSV file in which the first evaluation value and evaluation comments are written from the first generation AI 2.

[0028] Although the case where the interview execution prompt includes an instruction to create the first evaluation value and evaluation comment has been described, the present invention is not limited to this, and the first generation AI 2 may be instructed to conduct the interview and to create the first evaluation value and evaluation comment by separate prompts. The first generation AI 2 may also create the first evaluation value and evaluation comment after the interview is over, i.e., after answers to all questions have been obtained.

[0029] Although the case where a list of questions is provided to the first generation AI 2 in a CSV file has been described, the present invention is not limited to this, and the human resource development support device 1 may also provide individual questions sequentially as prompts as the interview progresses.

[0030] <1-3-4. Pass / Fail Judgment> FIG. 5A shows an example of training data used for model creation in the first embodiment. The training data includes, for each of a large number of employees identified by subject IDs, a second evaluation value obtained in a past interview and a pass / fail judgment result indicating whether each employee has reached the required level. Here, the second evaluation value is an evaluation value based on the same criteria as the first evaluation value described above. The pass / fail judgment result is a judgment result arrived at by a person such as a supervisor or a human resources consultant after carefully considering the second evaluation value and other circumstances, and obtaining such a judgment result requires considerable time and effort. Both the second evaluation value and the pass / fail judgment result are expressed as numerical values.

[0031] FIG. 5B shows an example of the assessment data used for score calculation in the first embodiment. The assessment data includes a first evaluation value obtained by the first generation AI2 for each employee identified by a subject ID. The first evaluation value is expressed as a numerical value. The employee identified by the subject ID in the assessment data may be a different employee from the employee identified by the subject ID in the learning data, and may be an employee of a different company if measures are taken, such as removing information that identifies the individual or the organization to which they belong, from the learning data. However, the assessment data does not include a pass / fail judgment result, and the pass / fail status of each employee in the assessment data is unknown information (the scores shown in FIG. 5B will be described later).

[0032] FIG. 5C shows details of the pass / fail determination module 15. The pass / fail determination module 15 includes a model creation module 151 and a score calculation module 152. The model creation module 151 uses the second evaluation value included in the training data as an explanatory variable and the pass / fail determination result as an objective variable to create a model showing the relationship between the explanatory variable and the objective variable. This model is given as a function showing the relationship between the explanatory variable and the objective variable. The score calculation module 152 calculates a score for each subject by inputting the first evaluation value of each subject included in the assessment data into the model. Specific scores are shown in the rightmost column of FIG. 5B. This score is a value corresponding to the objective variable in the training data and can be treated as a value indicating the probability of passing if the pass / fail determination is made using the same criteria as for the training data. In the present invention, the pass / fail of a subject who has undergone an interview can be determined based on a score calculated by comprehensively considering the evaluation values ​​of multiple elements in this way.

[0033] The pass / fail determination is not limited to the case where only the first and second evaluation values ​​are used. For example, the subject's past work performance, such as KPI (key performance indicator), may be acquired and included in the learning data and the determination data.

[0034] <1-4. Effects> According to the first embodiment, the human resource development assistance device 1 includes: an interview execution prompt providing module 13 that instructs the first generation AI2 to conduct an interview for talent evaluation and provides an interview execution prompt that instructs the first generation AI2 to create a first evaluation value of the subject based on answers from the subject in the interview; a model creation module 151 that creates a model showing a relationship between an explanatory variable and a dependent variable, using the second evaluation value of each of the plurality of learning objects as an explanatory variable and a pass / fail judgment result of each of the plurality of learning objects as a dependent variable; a score calculation module 152 that inputs the subject's first evaluation value into a model to calculate the subject's score; Includes:

[0035] According to this, the interview is conducted by the first generation AI2, allowing the subject to freely choose the time to proceed, and the first evaluation value can be evaluated objectively without human judgment. Also, the first evaluation value is input into the results of machine learning to calculate the subject's score, making it possible to determine pass / fail based on unified criteria, leading to appropriate human resource development.

[0036] According to the first embodiment, The interview is realized by presenting questions output by the first generation AI2 to the subject, accepting answers from the subject, and sending them to the first generation AI2. The human resource development assisting device 1 inputs at least a part of the first evaluation value into the model without presenting it to the subject.

[0037] According to this, in the interaction between the generation AI and the subject, instead of sending all of the output from the generation AI to the subject, at least a portion of the first evaluation value is not presented to the subject, so that the interview can proceed smoothly without the subject being overwhelmed with information or becoming happy or sad about each evaluation, and the human resource development support device 1 can obtain the evaluation value necessary for pass / fail judgment.

[0038] 2. Second embodiment <2-1. Functions and operations> Fig. 6 shows the functions and operations of a human resource development support device 1a according to the second embodiment. In addition to the various modules included in the human resource development support device 1 described with reference to Fig. 2, the human resource development support device 1a further includes a scenario creation prompt providing module 16, a training execution prompt providing module 17, and a diagnosis module 19. These modules are implemented by loading programs into memory included in the human resource development support device 1a and executing them with a CPU. The evaluation item creation prompt providing module 11, question creation prompt providing module 12, interview execution prompt providing module 13, and pass / fail determination module 15 are the same as those in the first embodiment.

[0039] The scenario creation prompt providing module 16 sets the subject who failed in the pass / fail judgment module 15 as a training participant, sets the "training purpose" and the "role of the trainer," provides a scenario creation prompt that instructs the subject to create a training scenario, and sends it to the second generation AI 2a. Even subjects who passed in the pass / fail judgment module 15 may be given training to overcome weaknesses, if they wish.

[0040] The training purpose and the trainer's role may be set as specified by a supervisor or a human resources consultant, or may be set by the human resources development support device 1a from among pre-prepared candidates based on the first evaluation value and evaluation comments. The scenario may be created by a supervisor or a human resources consultant, but since a human-created scenario may result in inconsistencies, it is preferable to have the second generation AI 2a create the scenario. The second generation AI 2a may be instructed to create a scenario taking into account the first evaluation value and evaluation comments so that the subject's weak points can be trained. Specific examples of the training purpose and the trainer's role will be described below with reference to Figures 8A, 9A, and 10A, and specific examples of the scenario will be described below with reference to Figures 8B, 9B, and 10B.

[0041] The training execution prompt providing module 17 provides a training execution prompt that assigns the role of trainer in the role-play to the second generation AI 2a and instructs it to create utterances in accordance with that role, and sends the training execution prompt to the second generation AI 2a. The training execution prompt instructs the second generation AI 2a to create utterances using the training purpose and trainer role set by the scenario creation prompt providing module 16 and the scenario received from the second generation AI 2a. A specific example of the training execution prompt will be described later with reference to FIG. 7. Training through role-play is conducted by the human resource development support device 1a mediating between the utterances output by the second generation AI 2a and the responses input to the subject terminal 3a. The subject terminal 3a has the same configuration as the subject terminal 3 and is a terminal operated by the trainee among the subjects who operated the subject terminal 3.

[0042] The training execution prompt may include an instruction instructing the second generation AI 2a to generate a third evaluation value and evaluation comments for the subject based on the subject's responses in the role-play. This instruction may be similar to the instruction included in the interview execution prompt. Alternatively, a prompt instructing the second generation AI 2a to generate the third evaluation value and evaluation comments may be sent separately from the training execution prompt.

[0043] The second generation AI2a that creates a scenario in response to a scenario creation prompt and the second generation AI2a that performs training in response to a training execution prompt may be separate generation AIs that include separate large-scale language models. The first generation AI2 and the second generation AI2a may be the same generation AI.

[0044] The diagnostic module 19 creates a diagnostic report on the subject's performance. The diagnostic report includes a third evaluation value and evaluation comments. Alternatively, the diagnostic module 19 may, like the interview execution prompt providing module 13, have the first generation AI 2 conduct another interview with the subject, causing the first generation AI 2 to generate a first evaluation value, and include a comparison of the first evaluation value before and after the training in the diagnostic report. Alternatively, the diagnostic module 19 may obtain the subject's work performance over a certain period after the training, such as KPIs, and include this in the diagnostic report. Furthermore, like the pass / fail determination module 15, the diagnostic module 19 may obtain learning data, perform machine learning to create a model, and input the third evaluation value, etc. into the model to calculate the subject's score. The diagnostic report can be viewed by a supervisor or a human resources consultant or sent to the subject's terminal 3, and used for human resource development.

[0045] <2-2. Specific examples> <2-2-1. Training Implementation> 7 shows an example of a training execution prompt for causing the second generated AI 2a to perform role-play training in the second embodiment. The training execution prompt includes a command statement instructing the second generated AI 2a to play the role of a trainer in accordance with the training purpose. The training execution prompt may also include a command statement pointing out any problems with the trainee subject's statements and instructing the trainee subject to return to the previous statement and resume the training.

[0046] <2-2-2. Example 1> Fig. 8A shows Example 1 in which "preventing power harassment" is set as the training purpose and "subordinate" is set as the trainer's role. Fig. 8B shows a scenario generated according to Example 1. Training is performed by providing the training purpose, trainer's role, and scenario to the second generated AI2a in addition to the training execution prompt.

[0047] FIG. 8C shows an example of training execution according to Example 1. The second generation AI 2a begins speaking as a trainer (subordinate). The human resource development support device 1a transmits the utterances output from the second generation AI 2a to the subject terminal 3a. The human resource development support device 1a receives the subject's response from the subject terminal 3a and transmits it to the second generation AI 2a. If there is a problem with the subject's response, the second generation AI 2a interrupts the role-play, points out the problem, and clearly indicates which utterance to resume from. Interrupting the role-play enables timely and effective feedback.

[0048] <2-2-3. Example 2> 9A shows Example 2 in which the training purpose is set to "improve management skills" and the trainer's role is set to "supervisor or customer." FIG. 9B shows a scenario generated according to Example 2.

[0049] Figure 9C shows an example of training execution according to Example 2. The second generated AI 2a starts speaking as a trainer (boss). To provide the subject with a training opportunity, the supervisor's speech may include ambiguous parts, thereby improving the subject's ability to cope with situations.

[0050] <2-2-4. Example 3> 10A shows Example 3 in which the training purpose is set to "improve customer satisfaction" and the trainer's role is set to "customer." FIG. 10B shows a scenario generated according to Example 3.

[0051] Figure 10C shows an example of training execution according to Example 3. The second generated AI 2a begins speaking as a trainer (customer). To provide the subject with a training opportunity, the customer's speech contains many questions at once, creating situations that are difficult to respond to, and improving the subject's ability to cope with the situation.

[0052] <2-3. Effects> According to the second embodiment, the human resource development assistance device 1a: a training execution prompt providing module (17) that provides training execution prompts that assign a role to the second generation AI (2a) and instruct the second generation AI (2a) to create utterances in accordance with the role; Human resource development training through role-playing is realized by presenting statements to the subject, accepting responses from the subject to the statements, and sending them to the second generation AI 2a.

[0053] This allows training after the pass / fail decision to be carried out using a generative AI, reducing the cost and effort of human resource development. Generative AI using large-scale language models has recently shown remarkable improvements in its ability to smoothly advance dialogue, enabling high-quality role-playing. Furthermore, since the human resource development support device 1a mediates between the utterances of the second generative AI 2a and the subject's responses, training execution prompts can be kept secret from the subject. [Explanation of symbols]

[0054] 1, 1a: Human resource development support device, 2: First generation AI, 2a: Second generation AI, 3, 3a: Subject terminal, 4: Database, 11: Evaluation item creation prompt providing module, 12: Question creation prompt providing module, 13: Interview execution prompt providing module, 15: Pass / fail judgment module, 16: Scenario creation prompt providing module, 17: Training execution prompt providing module, 19: Diagnosis module, 151: Model creation module, 152: Score calculation module

Claims

1. An interview execution prompt providing module that sends a list of questions and evaluation items to a first generation AI, instructs the first generation AI to conduct an interview for personnel evaluation according to the list of questions, and provides the first generation AI with an interview execution prompt that instructs the first generation AI to create a first evaluation value for the subject for the evaluation items based on the answers from the subject in the interview; an intermediary module that presents the question output by the first generation AI to the subject by transmitting it to a subject terminal operated by the subject, and mediates the interview by accepting the answer input by the subject from the subject terminal and transmitting it to the first generation AI; a model creation module that creates a model indicating a relationship between the explanatory variables and the objective variables, using the second evaluation values ​​of each of a plurality of learning objects as explanatory variables and the pass / fail judgment results of each of the plurality of learning objects as objective variables; a score calculation module that receives the first evaluation value of the subject from the first generation AI and inputs the first evaluation value into the model to calculate a score for the subject; A human resource development support device including:

2. 2. The human resource development support device according to claim 1, At least a portion of the first evaluation value is input to the model without being presented to the subject. Human resource development support device.

Images