34 of 37 Countries Expect AI to Take More Jobs Than It Creates: Inside Pew's 42,000-Person Global Survey

When Pew Research Center set out to map how the world feels about artificial intelligence, it asked 42,151 people across 37 countries a deceptively simple question: over the next 20 years, will AI mean more jobs or fewer?

The answer, published today, is unusually lopsided for cross-national survey research. In 34 of the 37 countries surveyed — from Australia and Germany to Nigeria and Indonesia — more people expect AI to subtract jobs from their economy than to add them. The report, led by researchers Laura Silver, Richard Wike, Moira Fagan and Jordan Lippert, is Pew’s most comprehensive global portrait of AI opinion to date, expanding on its 25-country survey from October 2025. Fieldwork ran from February 8 to May 13, 2026, with U.S. data drawn from two separate waves of the American Trends Panel.

A job-loss consensus, led by the rich world

The expectation of net job destruction is not evenly distributed — it tracks GDP per capita with striking consistency. In high-income nations, the pessimism approaches consensus: 76% of Australians and 76% of South Koreans say AI will result in job loss over the next two decades, followed closely by 71% of Americans — a figure that has climbed 7 percentage points in just two years in the only country where Pew has a prior benchmark.

In many middle-income countries, by contrast, the dominant response to the jobs question is not optimism but uncertainty. Roughly a fifth of adults or more are unsure how AI will affect their country’s labor market, and in some middle-income nations the unsure share reaches four in ten or higher. Pew’s analysts read this as opinion still taking shape rather than genuine optimism — the survey’s most optimistic finding is that many people simply haven’t decided yet.

Awareness cuts both ways

One of the report’s most counterintuitive findings concerns awareness. In about half of the countries surveyed, people who have heard or read “a lot” about AI are more likely than low-awareness respondents to predict job loss — familiarity, at least on the economic question, breeds pessimism.

Yet on overall sentiment, the pattern reverses. In Germany, among those who have heard a lot about AI, just 26% are primarily concerned and 15% are primarily excited; among Germans who have heard little or nothing, concern jumps to 41% while excitement collapses to 8%. Knowledge, in other words, simultaneously makes people more pessimistic about jobs and more comfortable with the technology overall.

A machine for inequality?

Asked whether AI will increase or decrease the gap between rich and poor, respondents again tilt darkly — and again the wealthy countries lead. In South Korea, Australia, the United States, Greece and Canada, four in ten adults or more say AI will widen inequality. Nowhere does more than a quarter of any country expect AI to narrow the gap, and in 29 nations that share sits in single digits.

The ideological divides are among the largest Pew measured anywhere in the report. American liberals are 31 percentage points more likely than conservatives to say AI will increase inequality. Age matters too: U.S. adults under 35 are far more likely than those 50 and older to expect widening inequality (56% vs. 38%). Gender adds another layer — in 11 mostly high-income countries, women are more likely than men to forecast job loss, with a 10-point gender gap in France.

The generational flip

Perhaps the most consequential trendline is generational. As recently as 2024, older people were reliably the most AI-wary cohort in the United States. That has inverted: the share of Americans ages 18–34 who are more concerned than excited about AI has jumped from 40% to 55% in two years, and young adults are now as likely as those 50 and older to be primarily concerned — a striking reversal from five years ago. Concern about AI-related job loss has risen fastest among precisely the age group expected to spend the longest career span working alongside these systems.

The same generational convergence is visible elsewhere. In Sweden, the gap in concern between the oldest and youngest adults has essentially disappeared within a year. Young adults in Poland, Brazil, Japan and Australia have also grown markedly more concerned.

Not despair — ambivalence

For all the economic gloom, the dominant global emotion toward AI is not fear but ambivalence. Across the 37 countries, a median 41% say they are equally concerned and excited about AI’s growing role in daily life, versus 37% primarily concerned and just 13% primarily excited. That mixed feeling is the most common response in about half of the countries surveyed.

Israel is the only country where the primarily-excited outnumber the primarily-concerned. South Korea offers the survey’s most fascinating paradox: South Koreans are among the very likeliest to forecast job loss and inequality (76% and roughly four in ten, respectively), yet 61% describe themselves as equally concerned and excited, and 21% — one of the higher shares anywhere — say they are primarily excited.

Concern is also compounding in much of Europe: the primarily-concerned share has risen since last year in Sweden (+9 points), Poland (+8), the Netherlands (+7), Hungary (+6) and Australia (+4).

America’s partisan flip

The U.S. data in the report — from a June 22–28 survey of 3,488 adults — documents a political realignment. Democrats are now more worried about AI than Republicans: 75% of Democrats expect fewer jobs over the next 20 years versus 68% of Republicans, and 56% of Democrats are more concerned than excited about the technology versus 49% of Republicans. That inverts 2023, when Republican concern led 59% to 46%. Democratic anxiety about AI-driven job loss has surged 17 points in two years.

Numbers captured before the summer of warnings

Timing matters when reading this survey. Fieldwork closed in mid-May — before Dario Amodei’s warning that AI could eliminate half of entry-level white-collar jobs, before the Hugging Face agent incident and OpenAI’s six-incident disclosure, and before the CEOs of OpenAI, Anthropic and Microsoft AI called for a development slowdown that President Trump dismissed as a “hoax.” As The Verge’s Thomas Ricker noted, the survey reflects a world polled “well ahead of recent apocalyptic warnings.” If anything, these figures likely represent a floor for public anxiety.

The report also updates a question with direct policy stakes: who should regulate AI. Pew’s 2025 predecessor found a median 53% across 25 countries trusted the EU to regulate AI effectively, versus 37% for the U.S. and 27% for China — a trust hierarchy that European Commission President Ursula von der Leyen is now explicitly courting with her planned Brussels summit of frontier labs.

With U.S. midterms seven weeks out, AI concern spiking among exactly the young and left-leaning voters Democrats need, and Congress already voting on AI’s physical footprint, the political economy of artificial intelligence is no longer a futurist debate. As this survey shows, the public made up its mind about the direction — the remaining argument is over who pays the costs and who captures the gains.