
Is AI taking jobs? In 2026 the answer is starting to show up in real data, not just forecasts. This page pulls together the most cited numbers on AI and the future of work, from the World Economic Forum, the IMF, Goldman Sachs and this year’s layoff reports, and lists the jobs most and least exposed. For background on the technology itself, see what is artificial intelligence.
Short answer: AI is taking some jobs, but so far it is changing far more than it removes. The IMF says about 40% of global jobs are exposed to AI, while the WEF expects a net gain of 78 million jobs by 2030. US companies blamed AI for about 120,000 announced cuts in early 2026. Hands-on, care and trade jobs are least exposed.
What do the big studies say about AI and jobs?
Short answer: high exposure, uncertain net loss.
| Source | Date | Headline finding |
|---|---|---|
| WEF Future of Jobs Report 2025 | Jan 2025 | 170M jobs created, 92M displaced by 2030; net +78M. Based on a survey of 1,000+ large employers |
| IMF (Kristalina Georgieva) | Jan 2024 | ~40% of global employment exposed to AI; ~60% in advanced economies, ~40% in emerging markets, ~26% in low-income countries |
| Goldman Sachs (Briggs and Kodnani) | 2023 | Generative AI could expose the equivalent of 300M full-time jobs to automation; could raise global GDP by 7% over 10 years |
| Challenger, Gray & Christmas | Oct 2026 | AI cited in 120,136 announced US job cuts in Jan–Sep 2026, ~21% of all cuts; top reason year to date |
| Microsoft Research (Tomlinson et al.) | Jul 2025 | Ranks occupations by how much their work overlaps with what Copilot users do with AI |
Sources linked below.
Two cautions. First, the WEF figure covers all forces reshaping work (green transition, demographics, economics), not only AI. Second, “exposed” is not “replaced”: the IMF notes that in many exposed jobs AI will complement workers rather than substitute for them.
Is AI already causing layoffs in 2026?
Short answer: companies say so, more than ever.
Challenger, Gray & Christmas, which tracks announced US job cuts, reports that employers cited AI for 120,136 cuts from January to September 2026, about 21% of the year’s 573,195 announced cuts. AI was the top reason given year to date. The monthly share has swung widely: in September AI was cited for 3,961 cuts, about 9% of that month.
Treat these numbers carefully. They count what companies say in announcements. Some cuts blamed on AI may be ordinary cost-cutting with a fashionable label, and some AI-driven changes, like not hiring replacements, never show up as layoffs at all.
Which jobs are most and least exposed to AI?
Short answer: language and information work is most exposed; physical and care work least.
The clearest occupation-level data comes from Microsoft Research’s 2025 study, which measured how closely each job’s work activities match what people actually use AI chatbots for.
| Most exposed to generative AI | Least exposed to generative AI |
|---|---|
| Interpreters and translators | Dredge operators |
| Historians | Bridge and lock tenders |
| Passenger attendants | Water treatment plant operators |
| Sales representatives of services | Foundry mold and coremakers |
| Writers and authors | Rail-track laying equipment operators |
| Customer service representatives | Pile driver operators |
| CNC tool programmers | Floor sanders and finishers |
| Telephone operators | Orderlies |
| Ticket agents and travel clerks | Motorboat operators |
| Broadcast announcers and radio DJs | Logging equipment operators |
Source: Tomlinson et al., “Working with AI” (Microsoft Research, 2025), Tables 3–4. The study measures task overlap, not job losses.
What jobs can AI not replace?
Short answer: jobs built on hands, presence and accountability.
The least-exposed list shares three traits, and they make a practical test for AI-proof jobs:
- Physical work in messy, changing environments. Plumbing, electrical work, construction, repair. Robots lag far behind chatbots.
- In-person care and trust. Nursing, elder care, therapy, teaching young children. People want a human, and someone must be physically there.
- Accountability. Someone has to sign off, take legal responsibility, or manage people. AI can draft the decision; a person still owns it.
If your job is mostly producing text, answering standard questions, translating or moving data between systems, expect it to change most. The usual advice is to learn to use the tools well and move toward the parts of the job that involve judgment and people.
Will AI create new jobs?
Short answer: yes, but not necessarily for the same people.
The WEF expects big growth in technology roles such as AI and big-data specialists, alongside frontline roles like care workers and farmworkers. Goldman Sachs pointed out that about 60% of today’s US workers are in occupations that did not exist in 1940, so new technology has historically created new work. The catch is timing and location: the new jobs may need different skills, in different places, from the people who lose out.
Could superintelligence change all of this?
Short answer: yes, which is why forecasts beyond 2030 are weak.
Every study above assumes AI keeps roughly today’s abilities. Labs are now openly aiming at AGI, systems that could do most cognitive work, and some leaders talk about super intelligence within years. If that happens, the “jobs AI can’t replace” list would shrink fast, starting with knowledge work and later physical work as robotics catches up. See our superintelligence timeline for when experts expect that, and will AI take over the world? for the larger risk debate.
Our take
The honest 2026 picture: AI is a real and growing reason for layoffs, especially in tech and customer-facing information work, but the economy-wide data does not yet show mass unemployment. The biggest mistake is reading “exposed” as “doomed”. The second biggest is assuming the next decade will look like the last two years. Plan for your job to change, and watch the history of AI: booms have surprised both optimists and pessimists before.
Enjoyed this? Get the SI Briefing
One short email a week: what changed on the road to super intelligence, and why it matters. No spam, unsubscribe anytime.
Frequently asked questions
Will AI take my job?
Probably not entirely, at least soon. Most studies measure ‘exposure’: the share of a job’s tasks AI can do. High exposure usually means the job changes. Roles made mostly of writing, translating, answering standard questions or data entry face the most risk of being cut.
What jobs can AI not replace?
Jobs that need hands-on physical work in unpredictable places, or in-person care: orderlies and nursing assistants, skilled trades, equipment operators, maintenance and repair work. Microsoft’s 2025 study found these had the lowest overlap with what generative AI does.
How many jobs will AI replace by 2030?
The World Economic Forum’s 2025 survey of employers expects 92 million jobs displaced and 170 million created by 2030, a net gain of 78 million. That covers all drivers, not only AI.
Is AI already causing layoffs?
Yes, according to what companies say. US employers cited AI in 120,136 announced job cuts from January to September 2026, about 21% of the total, per Challenger, Gray & Christmas. Some analysts think firms also use AI to explain cuts made for other reasons.
What are AI-proof jobs?
No job is fully AI-proof, but the safest combine physical skill, human trust and accountability: electricians, plumbers, nurses, care workers, equipment operators and roles that manage people or take legal responsibility for decisions.
6 Sources
- World Economic Forum: Future of Jobs Report 2025 press release (8 Jan 2025)
- IMF blog, Kristalina Georgieva: AI Will Transform the Global Economy (14 Jan 2024)
- Goldman Sachs: Generative AI could raise global GDP by 7% (2023)
- Challenger, Gray & Christmas: September 2026 job cuts report (1 Oct 2026)
- Challenger, Gray & Christmas: September 2026 report (PDF)
- Tomlinson et al. (Microsoft Research): Working with AI: Measuring the Applicability of Generative AI to Occupations (arXiv, Jul 2025)