The Ministry of Employment and Labor held the APEC Future of Jobs Forum for two days beginning April 6 at the Conrad Hotel in Seoul. As a follow-up to the joint statement of the APEC Labour and Employment Ministerial Meeting held in Korea last year, some 100 people attended, including international organizations, policymakers from member economies and private-sector officials. The discussions over the two days did not stop at how many jobs AI eliminates. The focus was on which group the shock reaches earliest.
Noh Se-ri, a researcher at the Korea Labor Institute, presented findings that most domestic firms put AI's rate of task replacement at up to 10%. She added that by OECD measures as well, no broad negative effect of AI on employment has emerged. A replacement rate of 10% is far from the fear that half of all jobs will vanish. Yet the latter part of the same presentation unsettles that reassurance.
After generative AI spread in 2023, growth in youth employment slowed noticeably in occupations with high AI exposure. AI exposure is an indicator of how much of a given job's tasks AI can currently perform. The rate at which people aged 22-25 enter these occupations is also declining. Total employment does not budge while only the door for new entrants narrows, a phenomenon the forum called the "aggregate trap."
Another assessment held that corporate workforce reductions are being made on the basis of expectations for AI in place of its demonstrated results. That means personnel plans are moving before the actual effects of adoption are verified. In such cases, those cut are generally workers with short careers who are seen as more easily replaceable. While aggregate statistics look calm, the burden concentrates on certain age brackets and skill levels.
Angelica Salvi del Pero of the OECD said in her first-day keynote that 27% of all occupations are exposed to contraction from automation. Findings were also presented that 54% of workers are concerned about AI-driven job loss over the next 10 years. High-level technical jobs that directly develop and maintain AI systems, meanwhile, account for only 1% of the total. The point is that the scale of the fear and the width of the newly opening channel do not match.
A similar gap appears in domestic surveys. In a survey by Gabjil 119, six out of 10 workers in their 20s said AI would take over their work, and the more unstable their employment or the lower their wages, the sooner they expected that point to arrive. The Korea Employment Information Service report "Digital Transformation and the Labor Market" analyzes that AI reconfigures jobs rather than simply reducing them. Within the same occupation, tasks that are automated and tasks that grow in importance diverge.
Corporate cases introduced at the forum show what that reconfiguration looks like. Hyosung ITX provided AI software training to call center agents and moved them into chatbot design and quality management work. POSCO replaced high-risk blast furnace inspections with a deep learning-based remote operation system, lowering the risk of industrial accidents, and the Occupational Safety and Health Council of Hong Kong presented a safety system combining smart helmet sensors with AI algorithms. How many workers these transitions applied to and how much accidents fell were not included in the presentations.
The obstacle to transition lies in capacity more than in will. Half of the small and medium-sized firms responding to the OECD survey cited technical capability as the biggest barrier to adopting AI. A presentation also noted that job postings for specialized AI software engineers rose 11% from the previous year, but the pool of people who can apply to those postings is limited. The gap between organizations with the capacity to redesign work and those without determines how hard the shock hits individuals.
Noh proposed building a lifelong learning system, expanding social dialogue and worker participation in the process of adopting technology, and responsible AI governance as alternatives. The government is preparing a framework plan on employment stability during industrial transition, linked to the national policy task of the "great labor transition," and its title and announcement date have not been set. The forum's discussion took up the roles of government with reference to the cases of Singapore and the Philippines.
AI does not empty out workplaces overnight. Instead it raises the threshold for entering a first job bit by bit, and quietly changes which tasks matter within the same job. The process by which a call center agent moves into a chatbot designer's seat now will determine the résumés of new entrants to society a few years from now. The calmer the statistics, the greater the need to look at what is shifting inside them.
