Jobs added in 2025 in occupations highly exposed to AI came to about 250,000. Between 80% and 90% of that increase came from Seoul, Gyeonggi and Incheon. That means more than eight of every ten went to the capital region. Over the same period, jobs heavily affected by robots and automation, such as machine operation and simple labor, were spread more thickly outside the capital region.
AI exposure measures how likely the tasks that make up a job are to be handled through AI or automated. High exposure cannot be read as meaning the work will soon disappear. New jobs also emerge on the high-exposure side. The 250,000 of 2025 are the evidence.
The dividing line was the type of AI. Occupations highly exposed to generative AI, which involve tasks such as writing documents and processing information, had the highest exposure in Seoul, followed by Sejong, Gyeonggi and Incheon. Physical AI, which carries out physical work with robots, showed high readings outside the capital region, in places such as Gyeongbuk, Jeonnam, Chungbuk and Ulsan. That is because industries where machines can directly replace human hands, such as manufacturing, transport, lodging and food service, are clustered in those regions.
Even within AI, the regions with high exposure differed by type.
The concentration is even greater for agentic AI, which carries out tasks on its own. Figures presented show that of the 105,000 jobs added in 2025 in occupations highly exposed to this field, 93,000, or 88.3%, were concentrated in the capital region. Whether these 105,000 are part of the overall 250,000 or counted separately cannot be determined from the explanation made public.
The results differed by age. Workers in their 20s declined nationwide in occupations highly exposed to AI. How large the decline was is not shown in this tally.
The Bank of Korea proposed applying AI to regional manufacturing for product development and process improvement. If the AI that enters factories stays confined to replacing people, regions outside the capital only lose. Attaching AI to areas where human experience has accumulated, such as design, quality control and defect prediction, creates work of a different character in the same factory.
Ulsan is a place worth watching to gauge that possibility. It ranks fourth among the country's 17 cities and provinces in labor market exposure to physical AI, and fifth in overall AI readiness. The outcome for a region with both ample room to introduce robots and the base to absorb them can be confirmed in employment statistics a few years from now.
The path along which the technology spread followed the industrial map exactly. Office functions that handle documents and information are concentrated in the capital region, so the work generative AI creates comes from there. Robots go where the machines and production lines are.
A process engineer in Ulsan and a planner in Seoul met the same technology in the same year, but one side got postings for new roles and the other got notices about the installation of automated equipment. Exposure works less as a tool for predicting who will be pushed out than as a sketch showing what work can be redesigned and where. When the destinations of the 250,000 are tallied again this time next year, the answer will depend on what was handed to AI inside the factories.
