Busan is good

Busan Metropolitan City Launches Nation’s First “Generative AI Administrative Service”

내용

Busan Metropolitan City (Mayor Park Heong-joon) announced that it has built the “Busan-style generative Artificial Intelligence (AI) administrative service” (hereinafter referred to as the “AI administrative service”) and will fully enter the operational phase starting next year (2026).

The core of this “AI administrative service” is to enhance the overall way administrative work is carried out by utilizing AI technology. Beyond digital administration, Busan Metropolitan City aims to systematically promote a transition to AI-based administration—improving administrative efficiency while accelerating responses to citizens. Rather than simply introducing a new system, the service is designed to redefine how administrative work is performed and what roles are involved, enabling AI to support or replace repetitive and standardized tasks previously handled by people.

To this end, Busan Metropolitan City has been promoting the “Busan-style AI Service Introduction Project” since April. After implementing a pilot service in some departments in October, the City held a completion briefing session on December 17 to comprehensively review the project’s outcomes.

As this project was carried out in an unprecedented setting, there were initial trials and errors as well as challenges in adapting to the technology. However, through this process, Busan Metropolitan City achieved meaningful results by accumulating practical experience and operational data necessary for introducing AI into administrative work.

The “AI administrative service” is the nation’s first generative AI administrative service built on a language model (sLLM)* and retrieval-augmented generation (RAG)** using internal materials. This has drawn strong expectations in that it provides a controllable operational environment suitable for AI use in the public sector.

* Language Model (sLLM, small Large Language Model): A language model optimized for specific tasks, capable of implementing stable and controllable AI services with relatively fewer resources.
** Retrieval-Augmented Generation (RAG): A method that generates responses based on relevant documents retrieved in advance, enhancing the accuracy and reliability of answers.

Through this approach, Busan Metropolitan City has strengthened the protection of internal administrative information and its data governance framework by operating AI primarily on internal administrative data without relying on external commercial generative AI tools. With this structure, the City plans to build a standard model that can be expanded across the broader public sector.

The “AI administrative service” provides 22 administrative functions based on internal administrative information, including Q&A, drafting document templates, and searching administrative materials. To ensure trust and verifiability in administrative use, all generated answers present the original source documents together. The service is delivered through a conversational interface linked to the NAVER Works collaboration tool.

In addition, by integrating AI functions into collaboration tools that civil servants already use daily, the City has minimized the learning burden associated with adopting a separate system. In particular, by offering various modes of use—including conversational interactions and memo-style formats—it is creating an environment where AI can be naturally utilized within day-to-day workflows.

Busan Metropolitan City will pursue the “AI administrative service” not as a short-term pilot project but as a mid- to long-term administrative innovation strategy. Over the next three years, the City will progressively advance the technology and expand the service, with the aim of establishing the Busan-style AI administrative service as the standard model for AI administrative services in the Republic of Korea.

A City official added that this project is “the starting point of a major AI transformation (AX),” shifting from “human-centered manual administration” to “AI- and robot-based automated administration.”

Cho Young-tae, Director General of the Administrative Management Bureau, emphasized, “This ‘AI administrative service’ goes beyond a simple technology introduction—it is a transition process to fundamentally improve the way administrative work is carried out.”

He continued, “We will establish a structure in which AI supports administrative work so that civil servants can focus more on essential judgment and planning tasks, and we will continue to pursue administrative innovation that can be felt not only within administrative operations but also by citizens, through safe and reliable use of AI.”