Friday, May 22, 2026

Gangnam Perspective: The Labor Market Even AI Doesn't Know

Input
2026-05-21 18:23:43
Updated
2026-05-21 18:23:43
Yunju Jo, Information and Media Desk
Since the spread of Generative Artificial Intelligence (Generative AI), the dominant frame in labor market discourse has been the idea of job extinction. The question most often asked has been, "Which jobs will disappear because of AI?" From lawyers and accountants to developers and designers, the view that AI will replace a large share of white-collar work has already become widely accepted. The sight of Generative AI writing reports, generating code, and summarizing contracts was undeniably startling, and fears of an era in which humans would no longer be needed quickly spread across society.
In that sense, the Qualitative Employment Outlook Report 2025–2035 recently released by the Korea Employment Information Service (KEIS) is highly interesting. Its outlook differs sharply from the familiar narrative of job extinction.
The report concludes that none of the 182 occupations analyzed are expected to disappear completely. Even the jobs identified as likely to decline were concentrated in highly repetitive and standardized areas, such as cashiers and design and editing assistants. By contrast, demand for workers in the medical, care, and cultural content sectors is expected to grow.
Of course, the report's forecast is far from certain. Given the pace of AI development, no one can confidently predict what the labor market will look like in 10 years. Generative AI has penetrated deeply into human work in just a few years since its debut. Functions that now look like mere support tools could still produce a far greater shock than expected.
Even so, the report matters because it raises different questions for a field that has tended to view AI only as a machine for eliminating jobs.
Another interesting point is that even AI models themselves cannot give a consistent answer to this issue. When U.S. researchers recently asked ChatGPT-5, Gemini 2.5, and Claude 4.5 which jobs were most likely to be replaced by AI, each model gave a different answer. In other words, AI itself is not certain about the future of the labor market.
In fact, when speaking with AI industry insiders and researchers, one often hears surprisingly skeptical views about the claim that "AI will completely replace human labor." The reason is clear: today's AI is not perfect.
Current AI can produce sophisticated documents and carry out analysis, but it also suffers from the problem of "hallucination," in which false information is presented as fact. In fields that carry responsibility, such as finance, law, and medicine, the need for human verification has only grown stronger. That is why the view from the field that "AI is not a technology that eliminates people, but rather a tool that allows one person to handle far more work" carries real persuasive power.
Changes seen in actual industries are also closer to a reorganization of tasks than to the deletion of jobs. Accountants use AI to analyze financial data, doctors receive support from AI in reading medical images, developers review code generated by AI, and designers rapidly turn out drafts with Generative AI. Human roles are shifting from simple production to setting direction, making judgments, and verifying results. Paradoxically, as AI becomes more advanced, the areas it cannot replace — contextual judgment, ethical responsibility, and emotional empathy — may become even more valuable. That does not mean there is no reason for concern. Even if AI does not immediately eliminate jobs themselves, the skills required within those jobs are changing rapidly. Mid-level roles centered on repetitive work are under structural pressure right now, and the ability to use AI well is already emerging as a core competitive advantage in the labor market.
In the end, the key question for the labor market in the AI era may not simply be which jobs will disappear. More important is how human roles will be redefined. And the labor market today may be searching for that answer in a far more complex way than expected.
yjjoe@fnnews.com Reporter