Most of the World Expects AI to Cost Jobs, Pew Finds

The Pew Research Center published its 2026 global read on artificial intelligence on Thursday, and the headline finding is close to unanimous: in 34 of the 37 publics it covers, more people expect AI to lead to fewer jobs than more jobs. The study is built on interviews with 42,151 people across 36 countries between 8 February and 13 May 2026, with United States figures drawn from two separate surveys of 5,119 and 3,488 adults.

The interesting result is not the direction but the gradient. Among the 18 high-income countries in the sample, a median of 55% expect AI to cost jobs over the next twenty years. Among the 18 middle-income countries, that falls to 36% — and the gap is not simply optimism in its place. A median of 34% in middle-income countries say they are unsure, against 22% in rich ones. The countries with the most exposure to the technology are the ones that have made up their minds about it.

The same split runs through the mood questions. A median of 40% in high-income countries say they are more concerned than excited about AI, against 14% who are more excited; in middle-income countries the split is 31% to 13%, with the balance sitting in the middle. Worry about jobs specifically reaches roughly 70% or more in Australia, South Korea and the United States. Across all 37 publics, a median of 37% describe themselves as more concerned than excited and 41% as equally both. Israel is the only public in the set where more people lean excited than concerned.

American attitudes moved fastest, and among the young fastest of all. The share of US adults who are primarily concerned rather than excited is up seven points in two years. Among Americans aged 18 to 34, it went from 40% in 2024 to 55% in 2026 — a fifteen-point swing in the cohort that uses these products most. Pew finds young adults more worried about AI-driven job losses than people over 50 in Canada, France, Singapore, Sweden, India, Indonesia and Malaysia as well.

Inequality tracks the same contour. A median of 35% in high-income countries say AI will widen the gap between rich and poor, against 22% in middle-income ones. The United States sits at 46%; Colombia at 13%. Awareness follows income too — a median of 50% in rich countries say they have heard or read a lot about AI, against 27% in middle-income ones, and 28% of people in middle-income countries say they have heard nothing at all, against 9% in rich ones. More than half of adults in Japan and Germany have heard a lot; in Sri Lanka and Bangladesh it is around one in ten or fewer.

The regulation question produces the survey’s least comfortable answer for Washington. Asked which of China, the United States or the European Union they trust to regulate AI, respondents in Bangladesh, Malaysia, Pakistan, Sri Lanka and the West Bank and East Jerusalem put China first. The Philippines is the only country surveyed where the United States is trusted more than both China and the EU. The EU generally does best in Europe, which is the one place it is least likely to change anything.

One caveat travels with all of it: fieldwork closed on 13 May, so none of these answers account for anything that has happened since. That matters more than usual in a year when the people building these systems have started arguing publicly about pace while Washington and Beijing have both declined to slow down. The public in most of the world had already reached a settled, pessimistic view about jobs months before that debate became loud — which suggests the argument playing out among lab chief executives is not one their customers are waiting on.