In fiscal 2024, we conducted technical verification and environmental improvements for the business use of generated AI
- Published:
Based on the "Priority Plan for the Realization of a Digital Society" (approved by the Cabinet on June 21, 2024), Digital Agency is conducting technical verification related to the business use of generated AI (mainly text generated AI) and developing secure environments for the use of generated AI. This initiative aims to promote the business use of generated AI with the aim of increasing the sophistication and efficiency of administrative work.
Background and Objectives of Initiatives
In fiscal 2024, we implemented the following two initiatives.
- Phase 1: AI Ideathon and Hackathon
- Purpose: Discovering Effective Use Cases of Generative AI from Business Issues in Administration
- Technical verification for business use of generated AI and development of usage environments (started in June 2024)
- Purpose: To develop a systematic approach to promoting the operational use of generated AI in public administration
Outline of technical verification
Purpose
By identifying current issues and examining solutions to each issue, we aimed to compile a systematic approach that contributes to the promotion of the business use of generated AI in public administration.
Validation Methods
The following key questions were set for Digital Agency staff members and technical verification was conducted.
- To what extent will the voluntary business use of generated AI be promoted by preparing environments where the latest generated AI can be used and Confidentiality class-2 information can be handled?
- To what extent will the use of generated AI in administrative work be promoted if accompanied by support from experts?
- What are the challenges and solutions to promote the business use of generated AI?
Results of technical verification
1. Benefits of Generated AI Usage Environments
The following results were achieved by preparing environments in which the latest generated AI can be used and Confidentiality 2 information can be handled, and by providing the latest generated AI utilization functions.
- As of February 2025, the average number of users per week reached 134, and the average number of API requests (* 1) reached 1,877.
Events for the use of generated APIs and the expansion of the range of information that can be handled are thought to have accelerated the number of users and AI requests. - Cumulative 85 Generative AI Apps Created
As a result of a survey of some users (117 people), 12 people (10%) developed their own generation AI application (* 2) and used it for personal use, and 8 of them shared the application with their team and used it.
- * 1 Number of API requests
- It refers to the number of times a single request is consumed when generating a sentence or generating an image using a generation AI.
- * 2 Generative AI application
- It refers to an application that can automatically generate sentences, images, and sounds using the generated AI.
2. Effects of Accompanying Support by Experts
Expert escort support was provided to some staff, resulting in 70 per cent of targeted teams being operational (including scheduled).
3. Issues and Solutions
Observed Challenges
- Lack of need to leverage generative AI
- Lack of resources to tackle generative AI exploitation
- Insufficient knowledge and functions for effective use of generated AI
Solution to the problem
- Establishment of an environment where Confidentiality class-2 information can be handled
- Construction of environments for using generated AI with necessary conditions and functions
- Provision and enhancement of educational content on generative AI
- Development of a support system by experts