Civil servants' AI tools take shape in government's new 'AI Lab'

One tool, "National Budget at a Glance," automatically collects and converts budget explanatory materials from central government agencies, then structures the data — project names, supervising ministries, responsible departments, budget figures and year-on-year changes — so users can search, filter and sort the information on a single web page without opening individual files.

Another, "Disaster Alert Map," addresses the difficulty of tracking emergency text messages sent across multiple regions by plotting them on an interactive map. Users can filter alerts by disaster type, search by region and query records by date range, all on one screen.

The Ministry of Interior and Safety said Wednesday that civil servant-developed AI tools closely tied to real administrative needs have been steadily registered with the "AI Government Lab," which has been in pilot operation since July 3, and are now undergoing testing and verification.

The AI Government Lab is an environment built to let civil servants directly in charge of specific tasks create prototypes using conversational coding, AI coding tools, open data and public APIs — with the aim of improving work efficiency and addressing public inconveniences.

Prototypes built in the lab are registered in a public code repository — the Public Development Output Repository, or Public GitLab — where they undergo further verification and refinement before being applied to actual administrative operations or converted into public-facing services.

As of Wednesday morning, 38 projects had been registered in the AI Government Lab by 31 contributors — roughly two AI-driven administrative innovation ideas submitted per day.

Once a project is uploaded, it takes just two to three minutes to generate a working prototype. Participants can also exchange feedback with one another after a project is registered.

The ministry is reviewing plans to expand the lab's concurrent user capacity from the current 200 to 3,000 next year, and to introduce a comprehensive incentive scheme for participants who submit outstanding projects.

"It is very encouraging that projects closely tied to people's daily lives and administrative work — such as budget data search, disaster information visualization and civil complaint contact guidance — have been appearing one after another from the very start of the pilot," said Hwang Gyu-cheol, director general for AI government at the Ministry of Interior and Safety. "We will actively support efforts to rapidly test and verify the ideas of civil servants who know the field best, and to spread proven best practices across the pan-government system so that improvements in administrative services are felt by the public."