Korea's missing 268,000: what the Bank of Korea's youth jobs number proves, and what it doesn't

The Bank of Korea has attached a figure to something hiring managers have been doing quietly for three years. In a report titled "Is AI Behind the Slump in Youth Employment? The Changing Career Ladder and How to Respond," the central bank found that jobs held by people aged 15 to 29 fell by 285,000 between June 2022 and June this year, and that 268,000 of that decline came in industries with high potential for AI use ^[1][2][3].

That is about 94 percent of the drop sitting in the AI-exposed part of the economy, leaving roughly 17,000 for everything else ^[6][7]. The window runs four years, or an average of about 71,000 youth jobs lost annually ^[5][8][9]. The sector detail is the part operators should read twice: employment fell sharply in information services, publishing, programming and professional services, the fields that relatively well-educated young people have been entering ^[4].

Now the caution. "High potential for AI use" is a measure of exposure, not a demonstration that software displaced a specific hire. The Seoul Economic Daily column reporting the figures makes that point itself, noting that companies scaling back open recruitment in favour of experienced hires is a long-running trend that predates generative AI, and arguing from field observation that AI is accelerating rather than originating it ^[10]. The same column reports that jobs held by people in their 50s rose over the period, without giving a figure ^[11]. A shift from junior intake toward experienced intake would produce exactly that pattern, with or without a model in the loop.

The operator consequence is a supply chain problem, not a headcount problem. Entry-level work was where people were paid to be slow: assembling materials, doing small tasks, being told to go back and fix things ^[12]. The column's account of the current market is that even entry-level postings now demand hands-on experience and a portfolio, with generative AI skill added as a qualification on top ^[13]. A firm that stops hiring juniors stops manufacturing seniors, and the invoice for that arrives in about five years, off the books of whoever made this year's budget.

The policy response so far mostly funds the classroom. Seoul's G3 planning committee has discussed a "Youth AI Ladder," and the city said it would provide AI licences and support AI workspaces plus job- and startup-training programmes ^[14]. According to the column, one question kept surfacing in that discussion: "Is giving someone the chance to use AI the same as giving them the chance to work in the age of AI?" ^[15] Its answer is no, and its proposal is to fund real work instead: actual projects, a fixed period alongside a senior colleague, feedback on output ^[16]. Separately, the presidential Economic, Social and Labor Council's special committee on population structure and jobs is weighing continued employment for older workers against youth entry, and the column's suggested settlement is pairing rather than splitting, with a skilled worker owning a project while a younger one handles AI-driven research, analysis and execution ^[17][18].

Watch whether Korean policy money moves from course seats to paid placements, since one is countable and the other is not. Watch the 50s series against the under-30 series in subsequent labour releases, because if both keep moving the same way, this is a story about who gets trained, not about what AI can do.