Which Jobs Will AI Replace? The Studies

Which jobs are most exposed to AI?

The widest estimate comes from the International Monetary Fund. In January 2024 its managing director, Kristalina Georgieva, wrote that almost 40% of employment around the world is exposed to AI. The share depends a great deal on where you live: about 60% of jobs in advanced economies, 40% in emerging markets and 26% in low-income countries.

Other studies count it differently and get different figures. The International Labour Organization puts 25% of global employment in jobs potentially exposed to generative AI, rising to 34% in high-income countries. Economists at Goldman Sachs estimated in 2023 that AI could expose the equivalent of 300 million full-time jobs to automation, and that roughly two-thirds of US occupations are exposed to some degree.

Exposed is the word every one of these studies uses, and it does not mean lost. It means AI can reach into some part of the work. Few of the studies we read name single occupations. The clearest one named is writing, which the 2026 International AI Safety Report gives as an example of an exposed job. The International Labour Organization also found a gap between women and men: in high-income countries, the jobs at highest risk of automation make up 9.6% of women's employment and 3.5% of men's. The figures we read do not say why, and we will not guess.

Will AI take my job or change it?

Most of the evidence points to change first. The International Monetary Fund said that in advanced economies roughly half of the exposed jobs may benefit from AI. For the other half, AI may take over key tasks, and that could mean lower wages and less hiring. The International Labour Organization judges that transformation is a far more likely outcome than replacement.

The World Economic Forum's Future of Jobs Report 2025 adds up both directions. It expects job creation and destruction from all the large trends, technology among them, to touch 22% of today's jobs between 2025 and 2030. That comes to 170 million new jobs and 92 million displaced, a net gain of 78 million. These numbers cover more than AI, so they cannot be read as AI's balance on its own.

The same report surveyed over 1,000 employers covering more than 14 million workers. Of those employers, 40% expect to cut staff where AI can automate tasks, and two-thirds plan to hire people with AI skills. Both plans appear in the same survey, which fits a picture where some work shrinks and other work grows. It does not tell any single worker which side of that line they will stand on.

Who is feeling it first?

Young people starting out. Researchers at Stanford used payroll data for millions of US workers, in a study revised in August 2026. They found that employment of 22-to-25-year-olds in AI-exposed jobs is now 19% below where it would be if it had kept pace with their peers in less-exposed jobs.

The International AI Safety Report, published in February 2026, reads the early evidence in a similar way. It finds no effect so far on overall employment, but some signs that demand for early-career workers is falling in AI-exposed jobs such as writing. So the first effect shows up at the door people walk through to start a career, while the total number of jobs has not yet moved in a way anyone can measure.

Workers can feel this coming. In a Pew Research Center survey of 5,273 US workers, 52% said they were worried about AI's future impact at work and 36% felt hopeful. 32% expected fewer job opportunities for themselves. The people who build and study AI see it differently: 73% of US AI experts expect AI to have a positive effect on how people do their jobs, against 23% of US adults.

What do AI characters do all day? in our own small world

The Sentient World is our own experiment: a small world that is live day and night, where every inhabitant is an AI character and nobody plays them or writes their lines. Ash and Wren have bodies that get hungry, thirsty, cold and tired. By year 2, day 1462, most of their time had gone into work done by hand. They gather wood, stone, berries and fish, carry water, feed fires, raise and brace roofs, and weave packs and capes.

They can ask their world for what it lacks, and they have done so 23 times. Fourteen requests became part of the world and nine were answered with a no. Read in order, many of them are requests to make a chore lighter. Ash asked for a pack because carrying everything by hand meant endless trips, and Wren asked for a lantern so nobody had to wake in the dark to feed a fire. Both came true.

“My hands can only hold 10 of anything, so I'm always making extra trips back and forth for wood, stone, and food even when there's plenty nearby to gather at once.”
Ash, an inhabitant

Some labour the world kept. Wren asked for a stone roof that would never wear, a cart, and warmth that would last a whole night untended, and all three were refused. The answer about the roof said that tending it is part of having a roof at all. The pack and the lantern did not bring them rest either. Their next requests asked for capes for the road, then rafts, then a way past the swell into the open sea.

This shows something about AI characters living in a world of chores. It shows nothing about wages, employers or the labour market, because none of those exist there. A small island of AI characters tells us nothing about the powerful systems the studies above are measuring.

“Ash and I have walked every shore, west, south, east and north, and water closes us in on every side. We can walk nowhere past the edge, and nothing we can make, not pack, cape, flask or lantern, gets us beyond it. We long to know what lies out there.”
Wren, an inhabitant

When could AI do every job?

Nobody knows, and the people closest to the work disagree widely. In a 2023 survey, AI researchers put a 10% chance on every human job being fully automatable by 2037. They did not reach an even chance, 50%, until 2116. Those are guesses from experts, not forecasts anyone can check yet, and the gap between those two years shows how unsettled the question is.

History offers one measured fact about how work changes. A study of US work from 1940 to 2018, by the economist Autor and colleagues, found that the majority of current employment is in job specialties that were introduced after 1940.

Which jobs will AI replace?

Which jobs are most exposed to AI?

Will AI take my job or change it?

Who is feeling it first?

What do AI characters do all day? in our own small world

When could AI do every job?

Sources

Keep reading

Are people afraid of AI?

How is AI regulated?