Pew Poll: 34 of 37 Countries Expect AI to Shrink Jobs

Jeff Bezos has been making a very different bet on artificial intelligence.

The Amazon founder said in June that AI would create a labor shortage rather than leave people without work. His argument was that better AI would lower the cost of turning ideas into products and services, creating more work for people instead of making them redundant.

A new Pew Research Center survey shows that public opinion is running in the other direction.

People in 34 of 37 countries surveyed said they expect AI to produce fewer jobs over the next 20 years than more jobs. Across the 37-country sample, the median share choosing “fewer jobs” was 46%, compared with 9% who expected more jobs. Another 13% said AI would make little difference to job numbers, and 25% were unsure.

That is not a small warning from a handful of tech workers. Pew surveyed 42,151 adults across 36 countries between February 8 and May 13, 2026, then used separate U.S. surveys covering 8,607 adults. The U.S. results were folded into the 37-country comparison.

America is among the most worried nations

Americans gave one of the strongest job-loss responses in the survey. Seventy-one percent said AI would lead to fewer jobs in the United States over the next 20 years. Only 5% expected more jobs, 10% saw little difference and 14% were unsure. South Korea and Australia were higher at 76%.

Pew also found that fear is not limited to older workers who may be watching automation reshape industries. In several countries, adults ages 18 to 34 were more likely than adults 50 and older to expect fewer jobs. That pattern appeared in countries including Canada, France, Singapore, Sweden, India, Indonesia and Malaysia.

Nigeria was the lone country where the share expecting AI to add jobs was higher than the share expecting fewer jobs, at 27% versus 26%. The survey also found unusually high uncertainty in Thailand and the Philippines, where “not sure” was the most common response.

Pakistan, which was included in the study, leaned toward job losses too. Forty-four percent expected fewer jobs, compared with 22% expecting more, 21% expecting little difference and 11% who were unsure.

Bezos sees more work. Workers see fewer openings.

Bezos’s forecast rests on a familiar economic idea: when technology makes one task cheaper, people find new things to build, new businesses appear and demand for human labor can rise.

That may still happen.

The trouble for that argument is what workers can see right now.

AI has already become part of corporate workforce decisions. Reuters reported in June that U.S. employers announced 97,006 job cuts in May, with AI linked to 40% of them, according to Challenger, Gray & Christmas. Reuters also reported that Amazon had trimmed roughly 30,000 corporate roles since late 2025, with AI efficiency gains part of the explanation.

Those figures do not prove that AI is the sole reason people are losing jobs. Companies cut staff for many reasons, and an employer saying AI is part of a restructuring is not the same as proving that a machine directly replaced every worker affected.

There is another part of the story that may explain why the Pew result feels closer to reality for many workers: jobs can disappear from the hiring pipeline before mass layoffs become visible.

A Stanford Digital Economy Lab study using payroll data through June 2026 found that employment among workers ages 22 to 25 in highly AI-exposed occupations was about 19% below the level the researchers estimate would have occurred if those workers had kept pace with peers in less-exposed occupations.

The researchers said the gap appears to be driven mainly by reduced hiring rather than increased separations. Their results also show weaker employment in occupations where AI is more likely to automate tasks, with less evidence of a decline in jobs where AI is used to assist workers.

That finding matters because a young person does not need to be fired by AI to feel its effect. A job that is never posted, a junior position that is removed from a team, or a company that decides one experienced employee with AI tools can cover work once assigned to several entry-level workers can change the market without producing a dramatic layoff headline.

The Pew survey is measuring expectations, not the final outcome

Pew is not saying that 34 countries will actually lose jobs because of AI. It asked people what they think will happen over the next two decades.

That distinction matters. The labor market is already showing signs of both automation and new AI-related hiring. Companies are building teams around AI, changing job requirements and creating new technical roles at the same time that other roles are being reduced.

Still, the direction of public sentiment is hard to miss.

People are watching companies spend heavily on AI and then announce workforce cuts. They are seeing entry-level opportunities tighten in some AI-exposed occupations. They are also hearing tech leaders make sharply different predictions about what comes next.

Bezos says AI can open up so much new economic activity that the problem could eventually be a shortage of workers, and even Sam Altman doesn't believe AI will reduce jobs.

The Pew respondents are looking at the same technology and asking a simpler question: if AI lets companies produce more with fewer people, where does the extra demand for workers come from?

Right now, the public answer is overwhelmingly that there will be fewer jobs.

That does not settle the 20-year debate. It does show why the job-loss argument is no longer just a futuristic fear. For part of the workforce, especially people trying to get their first serious job in AI-exposed fields, the change is already showing up in hiring decisions.

View and download the full study pdf here.