By Lee Jong-Wha
AI is advancing at a breathtaking pace, and so are fears about its impact on jobs, especially for young people entering the labour market. The tasks generative AI can increasingly perform—writing computer code, summarising documents, analysing data, and handling customer inquiries—are precisely those that often provide young workers with their first foothold in the workforce. But warnings by some AI-industry leaders that entry-level white-collar jobs will be largely wiped out within a few years run well ahead of the evidence. While AI is beginning to affect some entry-level jobs, its effects remain limited and uneven, and there are few signs of economy-wide displacement.
In the United States, for example, a recent study found no broad employment decline associated with AI exposure. What did fall was employment among 22–25-year-olds in highly AI-exposed occupations: it now stands 19% below where it would be if it had kept pace with the rate among young workers in less-exposed sectors. Weaker hiring, rather than layoffs, drove this shift.
But even this evidence should be interpreted cautiously. While entry-level hiring in AI-exposed occupations began weakening around 2022, before ChatGPT was released to the public, this probably did not reflect the rise of AI—which, at that point, had limited capabilities—so much as rising interest rates, post-pandemic adjustments, and remote work. Moreover, though the decline in entry-level hiring by tech firms since 2024 might be attributable to AI, demand for tech workers outside the industry remains strong. And where AI usage primarily complements workers, rather than substituting for human tasks, employment is flat or rising.
More broadly, recent US data do not point to a youth-employment crisis. The unemployment gap between 20–24-year-olds and the overall workforce is near a multi-decade low. In other words, not every job is exposed to AI, and even those that are do not automatically imply displacement.
This is even more true beyond the US. A recent World Bank study found that, in high-income countries, online job postings in occupations vulnerable to AI substitution declined by 5.8% relative to less-substitutable occupations since ChatGPT’s release. In middle- and low-income countries, the decline was statistically insignificant. Lower AI adoption, weaker digital infrastructure, and the larger share of manual work limit AI’s immediate potential for displacement in developing economies.
In Asia, AI’s labour-market effects differ sharply across countries. In South Korea, youth employment fell disproportionately in industries with high AI exposure in 2022–25, while employment among workers in their 50s increased. But much like in the US, it is impossible to determine the extent to which this can be attributed to AI rather than to other economic, demographic, and labour-market changes.
The ASEAN economies have some exposure to AI. The International Labour Organisation estimates that occupations with more than minimal exposure to generative AI accounted for 22.9% of ASEAN employment in 2025, but only 3.3% of the workforce was in jobs with the highest exposure. There are signs of weaker employment growth among young workers in some exposed occupations and entry-level jobs, but no broader deterioration in labour-market outcomes for this age cohort.
In China, rapid AI deployment appears to be creating displacement pressures for some workers, ranging from taxi drivers to white-collar professionals. This may well compound the challenges faced by young people entering the labour force, but China’s high youth unemployment predates the AI boom, reflecting slowing growth, private-sector weakness, and skills mismatches.
In India, by contrast, the risks are concentrated in the large information-technology and business-services sectors. While industry leaders dispute that AI is causing mass layoffs, there have been workforce reductions at major IT firms, and cross-country evidence suggests that economies specializing in digitally deliverable services may face greater employment pressure.
Japan’s situation is very different. With rapid population aging implying severe labour shortages, AI looks less like a threat than a solution. Generative AI use remains relatively limited, suggesting that expanding productive adoption and worker training may be more pressing than preventing displacement.
Across Asia, greater efforts must be made to ensure that young people have the tools they need to engage productively with AI. Updating education and training programmes is essential. Schools and universities must teach AI skills and digital literacy, as well as complementary skills that are harder to automate, such as analytical reasoning, judgment, communication, creativity, and teamwork. AI use must enhance learning, not substitute for it.
For those already in the workforce, firms must offer periodic re-training opportunities. As one OECD study showed, retraining, worker consultations, and clear workplace guidelines can go a long way toward harnessing AI-driven productivity gains. As entry-level tasks are automated, firms must also provide internships, apprenticeships, and other work-based learning opportunities for younger workers.
The challenge is to use AI to strengthen human capital, rather than short-circuit its accumulation. My recent research finds that greater AI exposure in South Korea is associated with improvements in adult numeracy, with some evidence of larger gains among younger workers.
But no amount of AI training can substitute for robust labour-market institutions and sound economic policies. The details will differ across countries, but in general, labour-market rules should make it easier for firms to hire and train young workers. And macroeconomic stability, investment, entrepreneurship, and competition remain essential to creating productive jobs.
AI will almost certainly transform Asia’s labour markets, and young workers may feel the effects first. But the shift might not be as dramatic as some industry leaders would have us believe. In any case, the best defence against AI-induced disruption lies in acquiring the skills, experience, and opportunities to use the technology well.
The writer is Professor of Economics at Korea University.