80% of AI-Driven Job Growth Concentrated in Metropolitan Area: "Rising Risks of Employment and Wage Gaps"

Bank of Korea Presents Findings at BOK Regional Economy Symposium

Human and Physical Resources Highly Concentrated in the Metropolitan Area; Gap in AI Readiness Also Evident
"Need to Strengthen Knowledge Services in Regional Manufacturing"
"Regional Hub Universities Should Be Developed as AI Hubs"

The Bank of Korea has analyzed that over the past three years since the emergence of generative artificial intelligence (AI), the gap in job opportunities between the Seoul metropolitan area and non-metropolitan regions has widened, especially in industries with high exposure to AI. Despite the rapid spread of AI technology, jobs in high-exposure occupations have actually increased, but this growth has been concentrated mainly in the metropolitan area and among highly skilled workers. The report points out that companies in the metropolitan area make more active use of AI compared to those in non-metropolitan regions, and that both human and physical IT resources are concentrated in the Seoul area. As a result, it is suggested that future disparities in employment and wages by region may become even more pronounced.

250,000 New Jobs in High-Exposure Generative AI Sectors Over Three Years... 80% Concentrated in the Metropolitan Area

On September 15, the Bank of Korea held the “BOK Regional Economy Symposium” in Daejeon and presented these findings under the theme “AI and Regional Labor Markets—Risks of Widening Gaps and New Opportunities Among Regions.” The report was authored by Minsoo Jeong, Head of the Regional Economic Survey Team, and Boseong Kim, Deputy Head, as well as researchers Yoojeong Jeong and Soobin Jang from the Bank of Korea’s Research Bureau.

The research team, for the first time in Korea, categorized AI into three types—generative, agentic, and physical—and measured exposure levels by occupation and region, analyzing the initial changes AI diffusion has brought to the labor market.

The results showed that, in the early stages of the spread of generative and agentic AI, jobs in high-exposure sectors—particularly those in the metropolitan area—have increased.

Since generative AI began spreading widely in 2023, employment in high-exposure generative AI jobs has grown by 247,000 over the past three years, with the metropolitan area accounting for 199,000 of these jobs, or 80.8%. It was confirmed that in the metropolitan area, a 0.1 rise in generative AI exposure correlates with a 1.0 percentage point rise in employment growth. However, in non-metropolitan areas, the correlation between these variables was not statistically significant.

For agentic AI high-exposure occupations, of the 105,000 increase in employment last year, 93,000 jobs (88.3% of the total) were concentrated in the metropolitan area.

The research team analyzed that both factors may be at play: generative and agentic AI are only partially automating certain tasks, and rather than replacing labor, are being used to enhance information processing capabilities and thereby expand markets. In contrast, the number of employees in high-exposure physical AI occupations, mainly physical labor, has continued to decline in both metropolitan and non-metropolitan areas since 2017, as these jobs were already more vulnerable to automation.

Team leader Jeong said, “The fact that job growth occurred mainly in the metropolitan area suggests that, for the same level of exposure, the more active adoption of AI in this region amplified its effects on market expansion and employment. Given that AI development is an extension of the IT sector, the concentration of IT-related human capital in the metropolitan area has also likely contributed significantly to regional disparities.”

Meanwhile, the report noted early signs of wage increases for high-exposure generative and agentic AI jobs centered around metropolitan employers. Despite the increase in job openings, companies in the metropolitan region still report a higher level of perceived labor shortages than those in non-metropolitan areas.

However, among individuals in their twenties, employment in high-exposure generative and agentic AI occupations declined in both metropolitan and non-metropolitan areas. Team leader Jeong commented, “Since university education still focuses on traditional knowledge and information accumulation, it remains very difficult for young graduates to acquire the professional skills and experience required, regardless of region.”

AI Concentration in the Metropolitan Area, Lagging Preparedness in Non-Metropolitan Regions... "Local Communities Must Actively Utilize AI"

The research team also pointed out that, when analyzing exposure levels by AI type, occupation, and region, there was a distinct difference between generative/agentic AI and physical AI.

By occupation, exposure to generative AI was highest in white-collar jobs with a high share of document creation and information processing tasks (such as accounting, bookkeeping, and marketing), sales roles (including telemarketing), and professional fields (like data and software specialists). Agentic AI, in comparison, had relatively higher exposure among management occupations (such as customer service managers and insurance or finance managers). On the other hand, exposure to physical AI was greatest in roles involving device or machinery operation and in simple manual labor.

By region, generative and agentic AI exposure was highest in Seoul, followed by Sejong and Gyeonggi Province, with all three ranking top for exposure level; the share of the metropolitan area was substantial. In contrast, physical AI exposure was higher in Gyeongbuk, Jeonnam, Chungbuk, and Ulsan in that order, indicating greater exposure in non-metropolitan areas. This reflects regional characteristics: the metropolitan area has a relatively larger share of white-collar, management, and sales jobs, as well as a high proportion of knowledge-intensive service industries, whereas non-metropolitan areas have a higher concentration of manufacturing and agriculture/fisheries jobs.

The report pointed out the possibility that the strong concentration of human and material resources related to AI and IT in a few regions such as Seoul and Gyeonggi could further widen regional labor market disparities in the future. Team leader Jeong noted, “Key elements of AI hardware, such as semiconductor manufacturing, as well as knowledge services with high AI exposure, are much more spatially concentrated in the metropolitan area than other industries. If this trend continues, regional divides may widen further.” He added, “This leads to capital investment becoming even more concentrated in the metropolitan area, which in turn attracts more technology talent, reinforcing the cyclical structure of agglomeration economies.”

Differences in regional preparedness for AI were also identified. Analyzing AI readiness across six indicators—including digital infrastructure, AI human resources, and AI utilization criteria—showed that Seoul and Gyeonggi (the metropolitan area) had overwhelmingly higher readiness scores. Sejong and Daejeon in the Chungcheong region also exhibited high preparedness, while Gangwon and Gyeongsang regions had relatively lower scores.

Nevertheless, the report mentioned that future AI utilization could present opportunities. For example, as models of AI-human collaboration develop, the concentration of talent and R&D in the metropolitan area could become less pronounced. Additionally, advances in physical AI could promote the transformation of regional manufacturing into knowledge services, thereby slowing the trend toward centralization of planning and research functions.

Team leader Jeong recommended, “To professionalize regional service industries, local governments, universities, and public institutions must collaborate to reinforce AI training systems within each region. At the same time, there needs to be support for convergence between manufacturing and knowledge services, as well as fostering startups.” He emphasized, “Investment should also be expanded so that regional hub universities can become AI hubs, bringing together research, information, and innovation capabilities at the local level.”

This content was produced with the assistance of AI translation services.