Gallup releases 37-country AI perception survey results

On September 23, Gallup released the results of an AI perception survey of adults in 37 countries. In 34 of the 37, positive feelings about AI outweighed negative ones; the exceptions were the United States, Egypt and the State of Palestine. At the other end, wealthy Western countries such as the United States, the Netherlands and Canada were grouped as the countries that use AI the most and worry about it the most.

The results are the first batch from research Gallup is conducting with Microsoft. According to Gallup, they cover the first 37 countries to complete fieldwork this year, and the study will ultimately cover 140 countries. The survey measures people's relationship with AI across five dimensions: emotional response, awareness, frequency of use, expected impact and trust. According to reporting, at least 1,000 people were interviewed in each country, with margins of error ranging from 2.2 to 4.9 percentage points depending on the country.

There is a wide gap between knowing about AI and having used it. Median awareness across the 37 countries was 81%, but the median share who had ever used AI was only 43%, and the median share who had never used it was 57%. In Singapore, 96% of adults are aware of AI, 79% have used it and 46% use it daily. In Malawi, awareness was 12%, ever-use 3% and daily use 1%.

Expectations run ahead of experience. In 25 countries, majorities of AI-aware adults said AI would make their own daily lives better, and in 32 countries optimists outnumbered pessimists. Views of the country as a whole were somewhat less bright: optimists led in 29 countries but formed an absolute majority in only 18. The share saying AI will mostly help people in their country was 93% in China, 91% in Vietnam, 77% in Singapore, 69% in Israel and 68% in Nigeria, and lowest in Bangladesh at 34%, Egypt at 35% and the United States at 36%.

Through a sociologist's lens, the most interesting finding is that expectations do not track use. In Nigeria, only 20% have ever used AI, yet 68% said it would help their country. Vietnam pairs 43% ever-use with 91% expectation. It reads as a sign that stories about a technology settle into a society before people experience it directly, and that the frame in which those stories are told shapes attitudes first.

Governments build that frame. According to reporting, Husan Chahal, an AI risk researcher at the National Bureau of Asian Research, noted that many governments, especially in Asia, have framed AI first and foremost as a tool for economic development and have been eager to drive uptake. "That doesn't necessarily mean they're unconcerned about catastrophic risks," Chahal said. "But those risks may not be as prominent in their policy priorities."

Expectation and trust do not move at the same speed either. The median share who trust AI completely or a lot to provide accurate information was 36%, and only four countries had a trusting majority: Israel at 67%, Vietnam at 59%, China at 58% and Nigeria at 52%. In 35 of 37 countries, more people expected AI to help people than trusted its accuracy. People see the potential but do not take the answers at face value.

Trust rises with use. In the United States, the share who trust AI's accuracy completely or a lot was 45% among daily users, 26% among weekly users, 15% among less frequent users and 8% among those who have never used AI. Gallup cautioned that the survey cannot establish whether trust drives use, use builds trust, or both.

Curiosity sits at the center of the emotions. The median share feeling at least one of curiosity, happiness or excitement about AI was 72%, compared with 41% feeling at least one of worry, sadness or anger. Positive emotions were most widespread in China at 90%, Singapore at 89% and Kenya at 87%. Curiosity was the most common emotion in 29 countries, with a median of 64%, well ahead of happiness at 39% and excitement at 34%. Among negative emotions, worry ranked first in 35 countries, with a median of 32%, above sadness and anger at 12% each.

This is where what Gallup calls the paradox of the worried West emerges. The seven Western high-income countries surveyed occupied ranks 1 through 6 and 8 on worry. The share saying AI makes them feel worried was 74% in the United States, 67% in the Netherlands, 64% in Canada, 63% in the United Kingdom, 61% in New Zealand, 60% in Ireland and 57% in Malta. The median share using AI daily in these countries is 29%, higher than the 17% across the other 30 countries. According to reporting, Gallup senior scientist Pablo Diego-Rosell said: "The most consistent finding that is emerging is that these cluster of attitudes that we see in Western countries is pretty unique." He explained that the worried West combines high adoption, high concern and a belief that AI may be good for the individual but less likely to be positive for their country.

Within the same country, heavier users worry less. In the United States, worry stood at 68% among daily users, 80% among less frequent users and 74% among those who have never used AI. In Canada, it rose from 54% among daily users to 72% among less frequent users and 69% among nonusers, and similar patterns appeared in the United Kingdom, Ireland and New Zealand. According to reporting, 32% of Americans have never used AI, and 23% use it daily, higher than Vietnam's 12%.

Western worry also overlaps with social scenes. METAL reported that while tech executives urged AI adoption inside the G20 venue in Chapel Hill, protests over data-center water and power continued outside. According to reporting, Edward Lim, Vietnam country director at a start-up accelerator connected to the National University of Singapore, said concerns are rising in Singapore as news of AI agents going rogue spreads, while in Vietnam, where human talent remains cost-competitive with AI agents, there is mostly excitement. "What strikes me is how open people in Vietnam are to trying new things," Lim said, adding, "The poll reflects what I see on the ground."

The Gallup release METAL reviewed came out the same week world leaders discussed AI at the United Nations General Assembly in New York. In its conclusion, the release says AI is not yet everywhere but opinions about it are already widespread, and that leaders will need to pay attention not only to who uses AI but to whether people trust it and how it makes them feel. From a sociologist's perspective, what this survey shows is that attitudes toward a technology are formed first by the stories and institutions around it, not by how much it is used. The next thing to watch, as the study expands to 140 countries, is whether the worried-West cluster holds, or whether other countries come to resemble it as their use grows.