While the fear that artificial intelligence will wipe out whole occupations has spread into policy debate, an empirical outlook has produced a different picture. In the "2025-2035 Qualitative Job Outlook Project Report," released three days ago by the Korea Employment Information Service (KEIS), not a single one of 182 occupations was classified as facing a "decrease" in employment by 2035. "Status quo" accounted for 114, or 62.6%, and "slight decrease" came to only 12 (6.6%). It is the first result of an outlook project being carried out over three years.
The report picked 205 occupations out of roughly 490 in Korea, then analyzed 182 after excluding 23 managerial occupations. The results were divided into five grades: decrease, slight decrease, status quo, slight increase and increase. Occupations classified on the growing side number 56, combining 47 in "slight increase" (25.8%) and 9 in "increase" (4.9%). It is closer to meaning that the label of the occupation itself mostly survives, while the work done inside it changes.
Where the growth and the decline fall is fairly clear. Health care, caregiving and living support were identified as the growing job groups. Demographic changes such as entry into a super-aged society, longer average life spans, rising chronic disease and the spread of single-person households were presented as grounds, and medical specialists, nurses and social workers were named as promising occupations. On the culture and content side, cartoonists and webtoon writers, film and record producers, travel product developers and accommodation service workers were also included in the growing job group.
The declining side is the clerical group, where repetitive and simple work is concentrated. Teller clerks, bank clerks, and design and editing support work were cited as jobs under heavy pressure to shrink. On the other hand, even within clerical work, demand was projected to expand for data analysis, digital finance and asset management, and business planning and marketing planning. Within the single block called clerical work, the jobs that survive and the jobs that are pushed out are diverging.
A separate report the KEIS issued last April found white-collar workers about 5.5% more likely than blue-collar workers to be replaced by AI. The published explanation of what the figure measures is short. Even so, read together with the conclusion of this 10-year outlook, the two reports point to the same place. Work done sitting at a desk is closer to the reach of automation.
Signals on the ground are moving faster than the outlook. In the ICT cluster of Pangyo, Gyeonggi Province, hiring has narrowed for junior developers with 1-3 years of experience, and an executive at a domestic IT company said new hiring demand for simple coding, translation and basic accounting bookkeeping is effectively close to zero. Industry assessments also say AI coding tools such as GitHub Copilot handle the simple work of three or four junior developers.
On the other side, labor shortages and rising wages are being observed at the same time in the plumbing, electrical, construction and facilities trades. Enrollment at U.S. vocational schools rose by double digits from a year earlier, and at the United Colleges Group in Britain, enrollment in engineering and construction courses grew about 10% over three years, while undergraduate enrollment fell back for the first time in a decade.
There is also a side in which AI itself creates this demand. As AI data centers increase, technical workers are needed to build and maintain power supply and cooling facilities. Nvidia Chief Executive Jensen Huang said that building AI factories requires an enormous number of electricians and plumbers, and Professor Geoffrey Hinton also named plumber as the occupation at lowest risk of replacement. The International Labour Organization analyzed in 2023 that generative AI has considerable room to complement existing jobs by automating some tasks rather than eliminating whole occupations.
By total volume, there are calculations that jobs will increase. The World Economic Forum projected in its 2025 report that 92 million jobs will disappear and 170 million new ones will be created by 2030. The scope of the estimate and the countries covered are hard to gauge from the published summary alone. Even if the total rises, the difficulty of this transition is that the people holding the 92 million jobs that vanish are not the same as the people who will fill the 170 million that appear.
The conflict has already surfaced. The Hyundai Motor labor union is publicly raising employment issues tied to the introduction of AI and robots and is clashing with management, and the possibility of putting the humanoid robot Atlas on production lines has also been raised. President Lee Jae-myung looked over that same robot himself on Feb. 27 at an investment agreement ceremony at the Saemangeum Convention Center in Gunsan, North Jeolla Province. One machine is carrying both the symbol of industrial promotion and the symbol of job insecurity.
Huang, 63, who stood at the podium at Carnegie Mellon University's commencement on the 10th, said AI is unlikely to replace you, but someone who uses AI better than you can. He gave the example that even if AI takes over reading radiology images, the role of the specialist who diagnoses and cares for patients will not go away. Once the outlook reports for the next two years come out, it will become possible to track which jobs moved to which grade.
Until then, there is one thing an individual can check. When you break down your own workday, which way does the ratio tilt between the pieces that can be handed to a machine and the pieces that are finished only when a person is there.
