Job Risk Estimates — 2026

What Percentage of Jobs Are at Risk From AI in 2026?

The range of estimates for how many jobs AI will displace is wide enough to be confusing: 14% per OECD task-level analysis, 47% per the original Oxford Martin School paper, and 300 million full-time equivalent jobs globally per Goldman Sachs. All three figures are credible. All three are measuring different things.

The key distinction is between jobs that are technically automatable and jobs that will actually be automated in a given timeframe — and between tasks AI can do and employers who choose to restructure around that capability. Understanding what each estimate actually counts is what makes the data usable.

Get My Free Fossil Score

What the research says

92M

jobs displaced by 2030

WEF Future of Jobs 2025

170M

new roles created by 2030

WEF Future of Jobs 2025

41%

of employers plan AI-driven headcount reductions

WEF 2025

55K

job cuts explicitly attributed to AI in 2024

Challenger, Gray and Christmas

Common questions

Why does the OECD say 14% when Oxford says 47%?

The difference comes down to unit of analysis. Oxford's Frey and Osborne paper from 2013 rated entire occupations as automatable or not, leading to a high aggregate figure. The OECD reanalysed the data at the task level, finding that most occupations contain a mix of automatable and non-automatable tasks. Even in high-risk occupations, some tasks require physical dexterity, social judgment, or creative problem-solving that is not currently automatable. Using task-level data dropped the estimated share of highly automatable jobs to 14%.

What does Goldman Sachs's 300 million figure measure?

Goldman Sachs's estimate of 300 million globally affected refers to full-time equivalent jobs where at least 50% of tasks are exposed to AI automation. This is a broader definition than the OECD's 'highly automatable' threshold and a different unit than the WEF's displacement projection. Goldman Sachs also projected that generative AI could raise global GDP by 7% over 10 years, suggesting a net positive economic outcome even as significant displacement occurs. The 300 million figure represents exposure, not inevitable unemployment.

Which percentage figure is most useful for thinking about my own career?

None of them, taken alone. Aggregate percentages tell you about the labour market as a whole, not about your specific role. What matters for your own assessment is the task composition of your actual job, the AI adoption velocity in your specific industry, and how quickly employers in your sector are restructuring their operations around AI. An occupation-level risk percentage is a starting point, not an answer.

Has any of the predicted displacement actually happened?

Yes, documented displacement has begun in specific categories. Challenger, Gray and Christmas tracked 55,000 job cuts explicitly attributed to AI in 2024 in the US alone. Klarna's announcement that AI handled the work of 700 agents, Duolingo reducing contractor reliance by 50%, and IBM's decision to pause hiring for 7,800 roles it expected AI to fill are documented examples. The current displacement rate is lower than the most pessimistic projections, but it is accelerating.

Is the 41% WEF employer figure consistent with these estimates?

The WEF figure of 41% of employers planning AI-driven headcount reductions is consistent with the range of academic estimates because it measures employer intent rather than total job exposure. Even if only 14-47% of jobs are technically automatable, employers do not need to eliminate those roles entirely to reduce headcount. They can maintain output with fewer workers by using AI to increase the productivity of those who remain. That productivity effect is what the 41% figure captures.

Free Assessment

Find out your personal risk in 4 minutes

Free Fossil Score assessment. No account required.

Get My Fossil Score