The US is one of only three countries where Gallup's new global AI survey found more worry than hope

Gallup and Microsoft published the first results of a planned 140-country study on global AI attitudes on Sept. 22, drawing on roughly 1,000 adults surveyed in each of 37 countries, with a margin of error ranging from ±2.2 to ±4.9 percentage points depending on the country. The headline finding is a genuine outlier: positive emotions about AI outweigh negative ones in 34 of the 37 countries surveyed so far, except for the United States, Egypt and the State of Palestine. In the US specifically, 74% of AI-aware adults say the technology makes them feel worried, and only 35% expect it to mostly help the country -- a minority view in the one country that leads the world in building the technology.

The gap is not small. In Singapore, 77% of adults expect AI to mostly help their country; in China, 93% do. Singapore also reports 96% AI awareness and 79% of adults having used AI at some point -- both far above the 37-country medians of 81% awareness and 43% ever-used. China's figure sits alongside 90% of its respondents reporting at least one positive emotion toward AI, against a 37-country median of 72%.

The more granular US data complicate the easy read that familiarity breeds comfort. Gallup broke American respondents out by how often they use AI, and worry does not fall in a straight line as use increases: 68% of daily users say AI worries them, versus 80% of less-frequent users and 74% of people who have never used it at all. Daily users are, in fact, the *least* worried of the three groups -- not the most. TechCrunch's write-up of the same data led with the fact that a majority of daily users are still worried, which is accurate; what that framing leaves out is that daily use is the group where worry is lowest, not highest, in Gallup's own numbers.

Trust in AI's outputs follows a similarly steep gradient. Nationally, 45% of US daily AI users say they trust its results “completely” or “a lot,” compared with just 8% of Americans who have never used it -- a 37-point gap between people who've tried the technology and people who haven't. China's overall trust figure, by contrast, sits at 58%, well above the 37-country median of 36%.

“An abstract, conceptual definition proved more confusing than helpful.” -- Gallup researchers Charles Lau and Alvin Nugroho, on why the survey asks about AI using concrete examples -- writing text, making images, answering questions -- instead of defining “artificial intelligence” up front

That line, from the researchers who designed the questionnaire, explains a choice that shapes every number above: rather than asking whether people have used “artificial intelligence” -- a term Gallup's own testing found more confusing than clarifying -- the survey names specific tools and common tasks, and only asks the deeper trust and emotion questions of people who say they're aware AI exists at all. The team also split what earlier drafts treated as a single hope-versus-worry choice into separate yes/no questions about specific emotions, after cognitive testing showed people hold more than one feeling about AI at once, and it split “daily” users into finer frequency bands after finding that occasional and heavy users “strongly differed in their patterns of use and sentiments toward AI.” (Gallup ran the same design in all 37 countries, so the frequency and emotion breakdowns are directly comparable across borders, not just within the US.)

One detail worth stating plainly: this is a Gallup-Microsoft partnership, not an independent academic study running on its own budget. That doesn't make the finding wrong -- Gallup discloses its full methodology, and a company with a stake in AI's public reception co-producing research that found *more negative* American sentiment than industry optimists might hope for is not, on its face, evidence of a thumb on the scale. But which organizations are behind a survey like this is exactly the kind of detail a reader weighing “Americans are uniquely anxious about AI” should be able to see, and most of the coverage this survey generated didn't mention it.

Those two headline facts -- positive emotions outweighing negative ones in 34 of 37 countries, and worry being the most common negative emotion in 35 of 37 -- sound contradictory until the survey design is clear: Gallup asked about each emotion separately, as a yes-or-no, rather than forcing people to pick a single overall feeling. A majority of people in most countries can and do say yes to both “AI makes me curious” and “AI makes me worried” in the same interview. The global median for feeling at least one positive emotion is 72%; the median for feeling worried is 32%. Both are true at once for a large share of respondents -- which is a more accurate picture of how people actually hold two feelings about a new technology than either a pure-optimism or pure-anxiety framing would suggest.

None of this settles whether American AI anxiety is a rational read of a technology still short on guardrails, or a cultural reflex in a country that also leads the world in AI investment. What Gallup's first wave does establish is that the anxiety is not evenly distributed, it is not simply a function of how much AI someone has touched, and it will keep being measured, country by country, for as long as the remaining 103 nations take to catch up.