Headlines about AI and job loss swing between "nothing to worry about" and "half of all jobs gone". Neither is what the main research sources say. This page sets out what four primary sources actually report, what kind of evidence each one is, and what it can and cannot tell you.
The short answer: so far, the evidence shows AI changing jobs more than eliminating them in large numbers, but some occupations are projected to shrink. Employers surveyed by the World Economic Forum expect more jobs to be created than displaced by 2030, across all major trends, not AI alone. The ILO finds about one in four workers worldwide in occupations with some generative AI exposure, and sees transformation as the most likely impact. The OECD found little evidence of significant negative employment effects from AI so far, while warning they may take time to appear. US projections show declines in specific, mostly routine, occupations.
Four sources, four kinds of evidence
| Source | What kind of evidence | Headline finding |
|---|---|---|
| WEF Future of Jobs Report 2025 | Survey of over 1,000 employers about what they expect, 2025 to 2030 | Expected creation of 170 million jobs and displacement of 92 million, a net gain of 78 million, from all structural trends combined |
| ILO Working Paper 140, 2025 | Task-by-task scoring of occupations for generative AI exposure | One in four workers worldwide in an occupation with some exposure; 3.3% of employment in the highest category |
| OECD Employment Outlook 2023 | Review of research plus surveys of employers and workers | Little evidence of significant negative employment effects from AI so far |
| US BLS Occupational Outlook Handbook | Official 10-year employment projections by occupation, 2025 to 2035 | Declines in some occupations where the BLS names AI or automation, growth in many others |
What employers expect to 2030 (WEF)
The WEF's Future of Jobs Report 2025 draws on a survey of over 1,000 employers representing more than 14 million workers. Extrapolating from their answers, it projects job creation equal to 14 percent of today's employment, 170 million jobs, offset by displacement of 8 percent, 92 million jobs, for net growth of 78 million.
Two cautions matter. First, these are employer expectations, not counts of jobs created or lost. Second, they cover all structural trends, including technology, the green transition, demographics and economic conditions, not AI alone. AI and information processing are the technology employers most expect to transform their business, with 86 percent expecting it to by 2030, but the report does not attribute the 92 million to AI. The WEF also expects clerical and secretarial roles to see the largest decline in absolute numbers.
Which jobs are exposed (ILO)
The ILO scores the tasks in each occupation for how far generative AI could do them. It finds one in four workers worldwide in an occupation with some generative AI exposure, and 3.3 percent of global employment in the highest category. It concludes that, as most occupations consist of tasks that require human input, transformation of jobs is the most likely impact, while warning that efficiency gains can still mean fewer workers are needed if jobs do not evolve. Exposure is not a count of job losses; it measures where change is most likely. Clerical work is the most exposed group.
What has happened so far (OECD)
The OECD Employment Outlook 2023 found that, based on research that mostly predates the latest wave of generative AI, there is little evidence of significant negative employment effects due to AI so far. It suggests this may be because adoption is still relatively low and because firms have preferred to adjust through slowed hiring, voluntary quits and retirement rather than dismissals, and warns that negative effects may take time to materialize. It also reports that occupations at highest risk of automation, taking AI into account, account for about 27 percent of employment, and that three in five workers surveyed were worried about losing their jobs entirely to AI in the next ten years. Its main finding is that, so far, AI has affected job quality more than job quantity; see AI and job quality.
Which US occupations are projected to shrink (BLS)
The BLS projects employment for 2025 to 2035. In several occupations it names AI or automation among the reasons for decline or slow growth:
| Occupation | Projected change 2025 to 2035 | Our guide |
|---|---|---|
| Claims adjusters, appraisers, examiners and investigators | Decline 6% | insurance |
| Customer service representatives | Decline 5% | customer service |
| Insurance underwriters | Decline 4% | insurance |
| Graphic designers | Decline 2% | graphic designers |
| Writers and authors | Little or no change | writers |
| Paralegals and legal assistants | Little or no change | paralegals |
At the same time, it projects growth for many other occupations, including some highly exposed ones, such as software developers and medical records specialists, because demand is rising. Projections are estimates and can be revised.
What this means for you
- Job loss is concentrated, not universal: the clearest pressure is on routine clerical, processing and high-volume content work.
- Slower hiring may come before layoffs: the OECD found firms preferring to adjust through hiring and attrition.
- Exposure is not destiny: demand, and how employers choose to use AI, matter as much as what AI can do.
- Your tasks matter more than your title: see will AI take my job? for how to check your own work, with guides for 26 occupations.
For what to do about it, see how to future-proof your career from AI.
Questions and answers
How many jobs will AI replace?
No primary source gives a reliable count for AI alone. Employers surveyed by the WEF expect 92 million jobs displaced and 170 million created by 2030 across all major trends, not AI alone, and these are expectations, not counts.
Has AI caused job losses so far?
The OECD Employment Outlook 2023 found little evidence of significant negative employment effects from AI so far, mostly based on research before the latest generative AI wave, and warned effects may take time to appear.
Which jobs are most at risk from AI?
Routine clerical and processing jobs are the most exposed in ILO data, and the WEF expects clerical and secretarial roles to decline most. US projections show declines for roles such as customer service representatives and claims adjusters.
Will AI create more jobs than it destroys?
Employers surveyed by the WEF expect net job growth of 78 million by 2030 across all trends combined. Whether AI on its own creates more jobs than it displaces is not settled by current data.
What does the ILO say about AI and jobs?
That one in four workers worldwide is in an occupation with some generative AI exposure, and that transformation of jobs is the most likely impact because most occupations include tasks needing human input.