Will AI take my job? What the hiring data actually shows

Probably not all of it, and probably not this year. The best data so far shows very few layoffs caused by AI. But it also shows something quieter: in the jobs AI is best at, companies are hiring fewer young people.

What kinds of study measure AI's effect on jobs?

Most AI job numbers come from one of two kinds of research, exposure studies and outcome studies, and they answer different questions.

• Exposure studies ask how many jobs contain tasks AI could do. They measure potential, not what has happened.
• Outcome studies look at payroll records, surveys and job counts to see what has actually changed.

Headline numbers are usually exposure numbers. The more useful evidence now is outcomes.

How many jobs are exposed to AI?

A lot, though mostly to be changed rather than replaced.

Both bodies say exposure does not mean loss. The IMF says roughly half the exposed jobs in advanced economies may benefit from AI. The ILO concludes that "transformation, not replacement, is the most likely outcome", and that its figures show potential exposure, not job losses. The ILO also found that in high-income countries, 9.6% of women's jobs are in the most exposed group, against 3.5% of men's.

Are companies laying people off because of AI?

Very few, so far.

The Federal Reserve Bank of New York surveys businesses in New York and northern New Jersey. Its August 2026 survey found:

Among firms using AI, just over a third of service firms and more than 20% of manufacturers said they were retraining workers. The researchers' headline: businesses are using AI to transform work, not cut jobs. This is one region's firms, not the whole country.

A study from Denmark points the same way. Economists Anders Humlum and Emilie Vestergaard linked large surveys of AI chatbot adoption to official labour records. Two years after ChatGPT launched, they found no effect on earnings or recorded hours, and could rule out effects larger than 2%.

Is AI hurting young workers?

In the most AI-exposed jobs, fewer young people are being hired, and the strongest sign of this comes from Stanford. Erik Brynjolfsson, Bharat Chandar and Ruyu Chen use payroll data from ADP covering millions of US workers, through June 2026. Their findings:

• There is no evidence of widespread job loss across the economy.
• Employment of workers aged 22 to 25 in the most AI-exposed jobs is about 19% below where it would be if it had kept pace with less-exposed peers. More experienced workers in the same jobs show no such gap.
• The gap has kept growing since they first reported it in August 2025.
• It comes mainly from hiring fewer young workers, not from firing them.
• The drop is concentrated where AI mainly replaces tasks. Where AI mainly helps workers, jobs are flat or rising.

The authors call these early, descriptive results, not proof that AI caused the change. Some of the trends started before generative AI.

What does the government expect?

The US Bureau of Labor Statistics projects total employment will grow 3.5%, or 5.9 million jobs, between 2025 and 2035. It expects office and administrative support jobs to fall 4.0%, a loss of 752,100 jobs, the most of any group, partly because of automation tools including AI. It expects sales jobs to fall 1.4%. Meanwhile data scientist jobs are projected to grow 34.6%. The BLS says the effects of AI are highly uncertain and that it applies AI adjustments conservatively.

How do people use AI at work?

In chat, people work with AI slightly more often than they hand it whole tasks, according to Anthropic's Economic Index, which looks at how people use its Claude models. In its January 2026 report, 52% of Claude.ai conversations were classed as augmentation, where a person works with the AI, and 45% as automation, where the AI does the task. Through the API, which businesses use to build AI into products, about three-quarters were automation. The report also found that 49% of jobs had seen AI used for at least a quarter of their tasks.

In Anthropic's June 2026 survey of users, which is not representative of all workers, 10% rated losing their own job as likely or very likely. Early-career workers said AI could do the largest share of their work, and were the most worried.

Our take

The question "will AI take my job" is less useful than "will AI change who gets hired". On today's evidence, the risk is highest for people starting out in jobs where AI can do the core tasks, such as office support. People already in those jobs, and people whose work AI helps rather than replaces, have seen little change so far. That could shift quickly as more firms adopt AI, so the hiring numbers are the ones to watch.