Globally, More People Expect AI to Cause Job Loss Than Growth, subtitled 37-country survey also finds growing awareness of AI – and growing concern in some places, is a Pew Research Center report published on September 17, 2026 by Laura Silver, Richard Wike, Moira Fagan and Jordan Lippert of the Center's Global Attitudes research team. It reports how adults in 37 countries view artificial intelligence: its expected effects on jobs and on the gap between rich and poor, whether its increased use in daily life makes people more concerned or more excited, how much people have heard about it, and, in 12 countries, how much they trust China, the United States, the European Union and their own country to regulate it. The report builds on Pew's 25-country survey of 2025 and on its U.S. series on AI attitudes, which in some items runs back to 2021.
The report's headline finding is that in 34 of the 37 countries surveyed, people tend to believe AI will lead to fewer jobs rather than more over the next 20 years, with pessimism about jobs and inequality most common in high-income countries. It also finds large shares unsure in many middle-income countries, mixed rather than purely negative feelings about AI's increased use, and concern rising among young adults in several countries.
Method
The international data come from Pew's Spring 2026 Global Attitudes Survey: 42,151 adults in 36 countries, interviewed February 8 to May 13, 2026 by telephone, face to face and online under the direction of Gallup, Langer Research Associates and Social Research Centre, using national samples. The 36 are Argentina, Australia, Bangladesh, Brazil, Canada, Chile, Colombia, France, Germany, Ghana, Greece, Hungary, India, Indonesia, Israel, Italy, Japan, Kenya, Malaysia, Mexico, the Netherlands, Nigeria, Pakistan, Peru, the Philippines, Poland, Singapore, South Africa, South Korea, Spain, Sri Lanka, Sweden, Thailand, Turkey, the United Kingdom, and the West Bank and East Jerusalem.
The United States, the 37th country, was surveyed separately through two waves of the Center's American Trends Panel, a probability-based panel recruited by address-based sampling and fielded by SSRS online and by live telephone. Wave 187 surveyed 5,119 adults on February 17–23, 2026 (survey response rate 87%, margin of error ±1.6 points) and supplies the U.S. awareness figures; Wave 195 surveyed 3,488 adults on June 22–28, 2026 (86%, ±1.8 points) and supplies the U.S. figures on jobs, inequality and concern. The cumulative response rate, accounting for recruitment nonresponse and attrition, is 3% for both waves. A figure described as a 37-country median therefore combines 36 international samples fielded in February–May with a U.S. sample fielded in June.
Some chapters use subsets. The comparison of high- and middle-income countries covers 18 of each, classified by World Bank income groups, and excludes the West Bank and East Jerusalem for lack of comparable GDP data; its international base is 41,113 adults in 35 countries. The regulation chapter covers 13,696 adults in 12 countries. Political ideology was measured in 28 countries. Education is standardized across countries using the International Standard Classification of Education.
Questions asked
- Jobs (Q57): "Thinking about (survey country) over the next 20 years, do you think that artificial intelligence will lead to …?" More jobs, fewer jobs, will not make much difference, not sure.
- Inequality (Q56): "Do you think artificial intelligence will …?" Increase the gap between the rich and the poor in (survey country), decrease it, not make much difference, not sure. The topline notes that the question does not specify a future period and that some translations use the future tense; a non-probability experiment among Spanish-speaking U.S. adults found the tense did not change the shares saying AI increases or decreases inequality.
- Concern and excitement (Q55): whether, overall, "the increased use of artificial intelligence in daily life" makes the respondent feel more excited than concerned, more concerned than excited, or equally concerned and excited.
- Awareness (Q54): "Artificial intelligence, also known as AI, is designed to learn tasks that humans typically do, for instance recognizing speech or pictures. Have you heard or read a lot, a little, or nothing at all about artificial intelligence?"
- Regulation (Q58, shortened): "How much trust do you have in each of the following countries and organizations to regulate the use of artificial intelligence effectively?" The United States, China, the European Union and the respondent's own country.
Country results
| Country | Fewer jobs | More jobs | Not sure (jobs) | AI will increase rich–poor gap | More concerned than excited | Equally concerned and excited | More excited than concerned | Heard or read a lot about AI |
|---|---|---|---|---|---|---|---|---|
| Australia | 76% | 4% | 9% | 52% | 53% | 36% | 12% | 53% |
| South Korea | 76% | 7% | 9% | 54% | 18% | 61% | 21% | 24% |
| U.S. | 71% | 5% | 14% | 46% | 52% | 37% | 9% | 48% |
| Greece | 68% | 4% | 13% | 42% | 49% | 40% | 9% | 36% |
| Canada | 61% | 6% | 23% | 41% | 50% | 41% | 8% | 47% |
| U.K. | 61% | 7% | 24% | 34% | 42% | 49% | 8% | 51% |
| France | 60% | 4% | 27% | 35% | 39% | 47% | 13% | 54% |
| Poland | 59% | 2% | 14% | 38% | 45% | 40% | 10% | 41% |
| Peru | 57% | 9% | 25% | 20% | 37% | 37% | 17% | 29% |
| Italy | 55% | 7% | 23% | 36% | 49% | 36% | 15% | 51% |
| Spain | 54% | 9% | 20% | 35% | 39% | 44% | 16% | 35% |
| Chile | 53% | 8% | 22% | 16% | 40% | 33% | 13% | 31% |
| Germany | 51% | 4% | 22% | 32% | 33% | 55% | 12% | 55% |
| Hungary | 50% | 6% | 27% | 25% | 39% | 41% | 17% | 49% |
| Japan | 49% | 8% | 20% | 35% | 30% | 53% | 14% | 56% |
| Netherlands | 48% | 5% | 23% | 35% | 41% | 40% | 16% | 52% |
| Sweden | 48% | 7% | 23% | 32% | 40% | 39% | 20% | 52% |
| Argentina | 47% | 8% | 33% | 23% | 36% | 43% | 11% | 26% |
| Singapore | 46% | 14% | 30% | 27% | 23% | 61% | 15% | 53% |
| Colombia | 45% | 10% | 38% | 13% | 41% | 38% | 15% | 30% |
| Pakistan | 44% | 22% | 11% | 39% | 34% | 24% | 25% | 24% |
| South Africa | 44% | 14% | 24% | 28% | 32% | 42% | 18% | 37% |
| Mexico | 44% | 9% | 36% | 13% | 38% | 42% | 13% | 23% |
| Israel | 42% | 15% | 25% | 38% | 19% | 35% | 29% | 31% |
| Brazil | 42% | 13% | 34% | 24% | 52% | 33% | 8% | 31% |
| Malaysia | 41% | 13% | 37% | 11% | 31% | 58% | 11% | 34% |
| Indonesia | 37% | 19% | 32% | 14% | 37% | 50% | 12% | 27% |
| Ghana | 36% | 29% | 20% | 23% | 24% | 23% | 27% | 24% |
| Kenya | 35% | 14% | 42% | 24% | 34% | 48% | 13% | 16% |
| W. Bank/E. Jerusalem | 33% | 13% | 35% | 29% | 45% | 39% | 7% | 30% |
| Thailand | 32% | 11% | 49% | 17% | 24% | 62% | 9% | 29% |
| Sri Lanka | 32% | 23% | 37% | 16% | 8% | 65% | 8% | 9% |
| Bangladesh | 30% | 27% | 22% | 30% | 9% | 47% | 10% | 4% |
| Turkey | 29% | 16% | 32% | 24% | 25% | 44% | 16% | 20% |
| India | 29% | 21% | 35% | 23% | 20% | 28% | 17% | 13% |
| Nigeria | 26% | 27% | 30% | 22% | 21% | 30% | 25% | 31% |
| Philippines | 26% | 14% | 54% | 9% | 31% | 57% | 8% | 41% |
| 37-country median | 46% | 9% | 25% | 28% | 37% | 41% | 13% | 31% |
Countries are ordered by the share expecting fewer jobs. Jobs, inequality and concern figures for the United States are from the June 2026 wave; the awareness figure is from the February 2026 wave. Shares who did not answer are not shown.
Jobs
Across the 37 countries, a median of 46% expect AI to lead to fewer jobs in their country over the next 20 years, 13% expect it not to make much difference, 9% expect more jobs and 25% are unsure. In many countries about a fifth of adults or more are unsure, reaching around four in ten or more in some middle-income countries; the unsure share is 54% in the Philippines and 49% in Thailand. Expectations of job loss are highest in Australia and South Korea (76% each), the United States (71%) and Greece (68%). In Nigeria, 27% expect more jobs and 26% fewer.
The United States is the only country in which Pew had asked the question before. The share of Americans expecting fewer jobs rose from 64% in August 2024 to 71% in June 2026, a 7-point increase over two years.
Expectations of job loss rise with national income. A median of 55% across 18 high-income countries expect fewer jobs, against 36% across 18 middle-income countries, where more are unsure (34% against 22%). The report gives a correlation of 0.60 between GDP per capita and the share expecting fewer jobs, with Singapore an outlier among wealthy countries; the belief is least common in the lower-GDP middle-income countries such as Kenya, Nigeria and the Philippines.
Within countries, young adults (18–34) are more likely than adults 50 and older to expect job loss in 13 countries, including Canada, France, Singapore and Sweden among high-income countries and India, Indonesia and Malaysia among middle-income ones. The gap is among the largest in the Philippines, where 36% of adults under 35 expect fewer jobs against 11% of those 50 and older. In the United States large majorities of all age groups expect job loss; adults 18–34 are more likely to say so than those 50 and older, a difference the report attributes to Americans 65 and older, and concern about job loss has risen most among 18- to 34-year-olds since 2024. Older people in these countries are not more likely to see AI as a job creator; they are more likely to be unsure. Women are more likely than men to expect fewer jobs in 11 mostly high-income countries (by 10 points in France). In many middle-income countries, higher-income and more educated adults are more likely to expect job loss, largely because lower-income and less educated adults more often say they are unsure. In around half the countries, people who have heard or read a lot about AI are more likely to expect job loss.
Inequality
On balance, more people expect AI to increase than to decrease the gap between the rich and the poor in their country: the 37-country medians are 28% increase, 26% not much difference, 5% decrease and 36% unsure. The shares expecting a wider gap are highest in South Korea (54%), Australia (52%), the United States (46%), Greece (42%) and Canada (41%), largely the same wealthy countries most pessimistic about jobs; the high-income median is 35% against 22% in middle-income countries. In no country do more than one in four expect AI to decrease inequality, and in 29 countries the share is in single digits.
People on the ideological left are more likely than those on the right to expect AI to widen the gap in 8 of the 28 countries where ideology was measured: the United States, Spain, Australia, France, Greece, Italy, the United Kingdom and Argentina. The U.S. difference is among the largest measured, with liberals 31 points more likely than conservatives to expect AI to increase inequality. Americans under 35 are more likely than those 50 and older to say so (56% against 38%), as are younger adults in Australia, Canada, Poland and Singapore; in Argentina, Chile, Hungary, Italy and Peru older adults are more likely to say so. In about a third of countries, more educated and higher-income adults are more likely to expect AI to widen the gap.
Concern and excitement
Across 37 countries, a median of 37% say the increased use of AI in daily life makes them more concerned than excited, 41% equally concerned and excited, and 13% more excited than concerned. Mixed feelings are the most common response in around half the countries. Israel is the only country in which more people are primarily excited (29%) than primarily concerned (19%). The highest shares primarily concerned are in Australia (53%), the United States (52%), Brazil (52%) and Canada (50%). South Korea combines among the highest expectations of job loss and inequality with one of the lowest shares primarily concerned (18%); 61% of South Koreans are equally concerned and excited and 21% primarily excited. Concern and excitement are much less closely related to GDP per capita than expectations about jobs or inequality: the high-income median primarily concerned is 40% against 31% in middle-income countries, and primarily excited 14% against 13%.
In many countries, people with more awareness of AI feel more optimistic about it. Among Germans who have heard a lot about AI, 26% are more concerned than excited, 59% equally and 15% more excited; among Germans who have heard a little or nothing, the figures are 41%, 51% and 8%. The awareness comparison is not reported for the United States, for which the report cites insufficient sample overlap between the two questions, nor for Bangladesh.
Across age groups, the median shares primarily concerned are 29% among adults 18–34, 34% among those 35–49 and 40% among those 50 and older, while adults under 35 are the most likely to be equally concerned and excited (48%) and primarily excited (18%). Adults 50 and older are more likely than those 18–34 to be primarily concerned in 15 countries. People with less education and lower household income are also more likely to express concern, and less likely to have high awareness.
Concern has risen in some countries. Since 2025 the share primarily concerned has increased in Sweden (+9 points), Poland (+8), the Netherlands (+7), Hungary (+6) and Australia (+4). In the United States views were relatively stable over the past year but concern is significantly higher than in 2021. The rise is concentrated among young adults in several countries. The share of Americans 18–34 more concerned than excited rose from 40% in 2024 to 55%, while remaining stable or falling among those 35–49 and 50 and older, so that Americans under 35 are now as likely as those 50 and older to be primarily concerned, where five years earlier older Americans were somewhat more concerned. In Sweden, 24% of adults under 35 were primarily concerned in 2025 against 36% of those 50 and older; in 2026 the figures were 45% and 43%. Concern among young adults also rose in Poland, Brazil, Japan and Australia, though older adults there remain more concerned. Concern among adults 50 and older rose in Poland, Sweden, France, Hungary and the Netherlands.
Awareness
Across the 37 countries, a median of 31% have heard or read a lot about AI, 42% a little and 21% nothing at all. The share hearing a lot ranges from more than half in Japan (56%) and Germany (55%) to 9% in Sri Lanka and 4% in Bangladesh. Awareness rose in 11 of the 25 countries also surveyed in 2025, with the largest increases in Nigeria (from 17% to 31%), Hungary (38% to 49%) and the United Kingdom (41% to 51%), followed by Australia, Brazil and Indonesia (+9 each), Poland and South Africa (+7 each) and the Netherlands, Sweden and Canada (+6 each); by income group the increases came in seven high-income and four middle-income countries. The share of Americans who have heard a lot has nearly doubled since Pew began asking in 2022, from 26% in December 2022 to 48% in February 2026.
Awareness is correlated with GDP per capita (correlation 0.74): 53% have heard a lot in Singapore, where GDP per capita is close to $100,000, against 4% in Bangladesh, at about $2,600. The high-income median is 50% against 27% in middle-income countries, and the share who have heard nothing at all is 9% against 28%. In all 36 countries where age can be compared, adults 18–34 are more likely than those 50 and older to have heard a lot, by 30 points or more in 19 countries; the largest gap is in Spain (67% against 16%), and the U.S. gap is 19 points (59% against 40%). Awareness among young adults rose 19 points in a year in both Hungary and Indonesia, in Indonesia with no change among adults 50 and older. In every country more educated adults are more aware; in nearly every country higher-income adults are (in Spain, 50% against 25%); and in around half the countries men are more aware than women.
Trust to regulate AI
In 12 countries, mostly middle-income, Pew asked how much people trust the United States, China, the European Union and their own country to regulate the use of AI effectively. The medians were 43% trusting China, 35% the United States and 34% the EU, and 56% trusting their own country.
| Country | China | United States | European Union | Own country |
|---|---|---|---|---|
| Pakistan | 87% | 28% | 33% | 82% |
| Ghana | 73% | 70% | 66% | 58% |
| Bangladesh | 62% | 46% | 56% | 78% |
| West Bank and East Jerusalem | 58% | 35% | 34% | 32% |
| Sri Lanka | 55% | 50% | 46% | 64% |
| Peru | 43% | 40% | 34% | 37% |
| Singapore | 42% | 28% | 38% | 68% |
| Malaysia | 38% | 18% | 24% | 50% |
| Colombia | 37% | 34% | 29% | 37% |
| Philippines | 36% | 53% | 44% | 60% |
| Chile | 35% | 33% | 32% | 41% |
| Thailand | 30% | 27% | 30% | 53% |
| 12-country median | 43% | 35% | 34% | 56% |
China is the most trusted of the three in five places: Bangladesh, Malaysia, Pakistan, Sri Lanka, and the West Bank and East Jerusalem. Singaporeans trust China and the EU most; in Chile and Thailand the three are trusted at about the same rate; in Ghana, Peru and Colombia the United States and China are trusted about equally. The Philippines is the only country surveyed where the United States is trusted more than both China and the EU. In half the countries people trust their own country more than any of the three. Trust in their own country to regulate AI is highest in Pakistan (82%) and Bangladesh (78%), and about four in ten or fewer in Chile, Colombia, Peru, and the West Bank and East Jerusalem.
Trust tracks general favorability: 57% of Peruvians with a favorable view of China trust it to regulate AI, against 26% of those with an unfavorable view, a pattern Pew also found in 2025. Trust in foreign regulators also rises with awareness of AI; 76% of Sri Lankans who have heard a lot about AI trust the EU, against 44% of those who have heard less, who are more often unsure. The 2026 results come from a different and mostly middle-income set of countries than Pew's 2025 survey of 25 countries, in which people tended to trust the EU most, followed by the United States and then China, and most trusted their own country. The report notes that in its separate 37-country analysis of views of China, people in middle-income countries tended to view China more positively than those in high-income countries.
Relation to other surveys
The U.S. figures come from the same two American Trends Panel waves as other 2026 Pew releases. The February wave is the basis of Americans and AI 2026: Chatbots, Smart Devices and Views on Impact, and the 48% who had heard a lot about AI appears in both. The June wave is the basis of From Diagnoses to Treatments, Why Americans Use AI Chatbots for Health and of Pew's August 18, 2026 short read, which reported the same 71% expecting fewer jobs (64% in 2024) and 55% of adults under 30 more concerned than excited against 39% in 2024 (Source: pewresearch.org). The short read's under-30 group and this report's 18–34 group are different age bands, so the 39% and 40% baselines are not a discrepancy. Pew's July 23, 2026 release found 51% of Americans expecting AI to widen the gap between rich and poor countries (Source: pewresearch.org); this report's inequality item asks about the gap between rich and poor within the respondent's own country, on which 46% of Americans expect a wider gap.
The U.S. job-loss figure is close to the 70% of adults in the June 2026 Gallup–SCSP wave who expect AI to destroy more jobs than it creates, although Gallup–SCSP asks about a ten-year horizon and Pew about twenty years (American AI Sentiment in 2026 (Gallup and SCSP)). Gallup–SCSP concludes that among U.S. adults "higher use of and knowledge about AI tools are not associated with more positive sentiment toward it," while this report finds that in many of the countries surveyed, people with more awareness of AI feel more optimistic. The two findings do not directly conflict: Pew's awareness comparison excludes the United States, and the surveys measure different things (self-reported awareness against concern and excitement in Pew; use, knowledge, trust and economic expectations in Gallup–SCSP).
Provenance
The report was captured on September 18, 2026 from pewresearch.org, the Center's own domain: the web landing page, the report PDF (all five chapters, the appendix of detailed tables, acknowledgments, methodology and the topline) and the separately published topline PDF (Sources: pewresearch.org; pewresearch.org; pewresearch.org). Page metadata gives publication at 9:59 a.m. Eastern on September 17, 2026. The country tables on this page are taken from the report's main tables and checked against the topline, with which they agree. Several chart-derived tables in the PDF text extraction are garbled, including the age breakdowns of the jobs and concern items and the ideology chart; figures from those charts are given here only where the report states them in prose. Coverage on the day of release by Gizmodo reported the same 37-country medians on jobs (Source: gizmodo.com). USA Today's account gave the combined sample as 50,838 people including 8,687 in the United States (Source: usatoday.com); the report's own figures sum to 50,758 and 8,607.
Relationships
- supports: Public Opinion on AI — the first cross-national survey of AI attitudes of this scale held in the vault.
- supports: AI Labor Disruption — in 34 of 37 countries people tend to expect fewer rather than more jobs from AI.
- related: AI Political Cleavages — left–right differences on AI and inequality in 8 of 28 countries, largest in the United States.
- related: US-China AI Competition: Different Races, Different Metrics — trust in China, the U.S. and the EU to regulate AI in 12 countries.
- related: Americans and AI 2026: Chatbots, Smart Devices and Views on Impact, From Diagnoses to Treatments, Why Americans Use AI Chatbots for Health — Pew reports drawn from the same U.S. panel waves.
- related: American AI Sentiment in 2026 (Gallup and SCSP) — the U.S. survey series with comparable job-loss and sentiment measures.
- related: Pew Research Center — publisher.