AI adoption is often treated as a proxy for acceptance. A new Gallup survey suggests that is too simple.
Among Americans who use AI every day, 68% say they are worried about it, according to results reported by TechCrunch. Across the U.S. sample, 74% reported worry, while only 36% said they expect AI to mostly help the country. The finding matters because it separates personal utility—using a tool for research, presentations or routine tasks—from confidence in its broader effects on jobs, institutions, privacy and safety.
What the survey found
Gallup conducted the study for Microsoft between April and July, polling roughly 1,000 people in each of 37 countries. The project is ultimately intended to cover 140 countries. The reported group does not yet include India, Australia or Malaysia, an important limitation when reading the results as a global snapshot.
The U.S. sits at the anxious end of the reported results. Daily users are not uniformly convinced AI outputs are reliable either: just 45% of American daily users said they trust AI’s results a lot. The median across surveyed countries was 36%.
That pattern contrasts with several high-adoption markets. Singapore had the highest reported share of daily AI users, at 46%; more than 80% of people aware of AI there expected it to improve their everyday lives, and 77% thought it would help the country. Majorities in China and Israel also expected national benefits. Globally, Gallup said respondents in 34 of 37 countries most commonly described themselves as curious about AI, while 32% said they were worried.
The point is not that one sentiment has displaced another. Gallup senior scientist Pablo Diego-Rosell described attitudes as multidimensional: people can use AI frequently, expect benefits and still be concerned—or see its potential while doubting its accuracy.
Why operators should care
For companies deploying AI, usage metrics alone are an incomplete measure of product-market fit or organizational readiness. Employees may turn to AI because it saves time, while simultaneously fearing poor oversight, incorrect outputs, job displacement or misuse of their data.
That gap can show up as inconsistent adoption, reluctance to use AI in consequential workflows, or resistance to deployments in hiring, education, healthcare and public services. It also means a mandate to “use AI” without clear boundaries may not build durable trust.
A more useful operating approach is to distinguish between low-risk assistance and high-stakes decisions. Teams should specify where AI is permitted, what information cannot be entered into tools, when human review is required, how outputs are checked, and who is accountable when something goes wrong. Leaders also need to explain whether AI changes roles, performance expectations or staffing plans rather than leaving employees to infer the answer.
For AI vendors, reliability claims and product demos will not be enough. Buyers will increasingly look for auditability, controls, data handling commitments and evidence that systems perform acceptably in the specific workflows they are asked to support.
What to watch next
The survey does not establish why attitudes vary by country, and reported sentiment can shift quickly as AI capabilities, labor markets and policy debates change. Still, it signals that exposure by itself is unlikely to resolve public unease.
Watch whether businesses and policymakers respond with practical safeguards rather than assuming adoption will create legitimacy. The companies best positioned to sustain AI use may be those that make the trade-offs visible: where automation helps, where human judgment remains necessary, and how they will manage the risks users already say they see.
