ILO warns generative AI job shifts

Generative AI is turning into a labor-market story, not just a tech story. That matters because the real disruption is less “robots take all the jobs” and more “a lot of jobs quietly change shape.” The ILO’s Asia-Pacific team put that into focus on April 21, with a regional look at ASEAN that says exposure is real, uneven, and heavily concentrated in specific kinds of work. The headline is simple — most jobs are more likely to be reworked than erased. (ilo.org) ### What did the ILO actually say? The new ASEAN analysis says 21% to 28% of jobs in countries with comparable data are exposed to generative AI in some way. But exposure does not mean replacement. The ILO’s point is that most of these roles include tasks GenAI can assist with, not entire jobs that can vanish overnight. That sounds s(ilo.org) if the worker stays. (ilo.org) ### Which countries look most exposed? The Philippines sits near the top of the ASEAN range, largely because its economy leans hard into services and IT-BPM work. Thailand and Viet Nam come in somewhat lower overall. In Viet Nam, exposure is also geographically concentrated — Hà Nội, Ho Chi Minh City, and Đà Nẵng account for more tha(ilo.org)and hardest. (ilo.org) ### Which jobs are in the blast zone? Clerical support work is the clearest pressure point. In the Philippines, 93.7% of clerical roles are exposed to GenAI, and 37.8% sit in the highest-risk category. In Viet Nam, nearly two thirds of clerical support workers are in occupations most susceptible to GenAI-driven task automation. Basic(ilo.org) to matter. (ilo.org) ### So are layoffs the main risk? Not yet — and maybe not mainly. The ILO keeps stressing that the bigger effect is task transformation inside jobs. In Viet Nam, around 11.5 million workers — one in five — are in occupations with some GenAI exposure, but only about one million workers are in roles where tasks are both highly suscepti(ilo.org) “not eliminated” does not mean “not disrupted.” (ilo.org) ### Why does gender keep coming up? Because the exposure is not neutral. The ILO says women are more likely than men to be in jobs touched by GenAI, both globally and in country studies like Viet Nam. In Viet Nam, women’s exposure is 24.1% versus 17.8% for men. Across countries with available data, female-dominated occupations are almo(ilo.org)exposure bands versus 3% for male-dominated ones. That is mostly about occupational sorting — clerical, admin, payroll, reception, business support. (ilo.org) ### What are governments doing with that? Singapore is a good example of the policy turn. SkillsFuture says it will launch a self-diagnostic AI readiness tool by 2Q 2026, steer workers toward relevant courses, and work with industry partners to improve AI awareness and adoption among SMEs. It is also revamping the MySkillsFuture portal(ilo.org)tract AI hype and toward practical retraining, job redesign, and worker-side adoption. (skillsfuture.gov.sg) ### Why does this matter beyond ASEAN? Because ASEAN is a preview of the broader pattern. GenAI hits office-heavy, service-heavy, documentation-heavy work first. It does not arrive evenly. It does not hit all workers the same way. And it creates a strange split: the same tools that can raise productivity can also hollow out entry-level routine work if firms use them badly. The region’s big (skillsfuture.gov.sg)rs get redesigned jobs, better tools, and training fast enough to keep up. (ilo.org) ### Bottom line The ILO is basically warning against two lazy takes at once — “AI changes nothing” and “AI destroys everything.” In ASEAN, the more believable story is uneven task churn. That is messier, but more useful. It means the winners are likely to be the countries and employers that treat AI as a work redesign problem, not just a software rollout. (ilo.org)