On 12 March 2026, Pew Research Center published a five-year roundup of how Americans view artificial intelligence — 13 findings drawn from its surveys since 2021, written by Michelle Faverio and Emma Kikuchi as an update to the same series from November 2023. The headline number: 50% of U.S. adults say the growing role of AI in daily life leaves them more concerned than excited. Only 10% are more excited than concerned; 38% feel both equally.
The value of this roundup is longitudinal. A single monthly poll captures a mood; five years of data capture a trend — and the trend is unambiguous: usage and familiarity keep climbing, concern climbs with them, and excitement does not. For teams busy stuffing AI into every corner of a product, these numbers are the most direct market thermometer available.
Concern Up 13 Points in Five Years
When Pew first asked the question in 2021, 37% of U.S. adults were more concerned than excited; by the June 2025 survey that had reached 50%, and concern in the U.S. runs higher than in most of the 24 other countries Pew has polled. Expectations for society are equally cool: about half of adults say AI will make creative thinking worse and make it harder for people to build meaningful relationships. The few bright spots cluster in medicine — 44% expect a positive impact on medical care versus 19% negative — while education (24% positive) and jobs (23% positive) rank far lower, with another 16% to 30% unsure across areas. Familiarity has surged regardless: 47% of adults have heard “a lot” about AI, up 21 points from 26% in 2022 — and knowing more has not made anyone feel better.
Teens and Kids Are Already Deep In
A fall 2025 survey found 64% of U.S. teens ages 13 to 17 have used an AI chatbot. More than half use them for information searches and schoolwork help, and about one in ten rely on them for all or most of their schoolwork. The use breakdown: summarising text or video 42%, creating or editing images and video 38%, casual conversation 16%, emotional support or advice 12%. Roughly six in ten teens say chatbot cheating happens at least somewhat often at their school, with about a third calling it extremely or very often. Younger children are entering too: parents report 15% chatbot use among 11- to 12-year-olds, voice-assistant use at 53% in that age band and 12% even under age two, plus smartwatch use at 21% among parents of 11- to 12-year-olds and 11% across all parents of children 12 or under.
Workplace Adoption: Slow but Real
The share of U.S. workers who say at least some of their work is done with AI rose from 16% in 2024 to 21% by September 2025; on the other side, 65% use it little or not at all. Daily contact density is rising too: 31% of adults interact with AI several times a day, up from 22% in February 2024. The age gap is stark — ChatGPT use at work sits at 38% for employed adults under 30, 30% for ages 30 to 49, and 18% for those 50 and older, while roughly half of adults under 50 interact with AI daily or more. In other words, AI is already background infrastructure for half of Americans under 50, but formal adoption inside job workflows still trails personal use by a wide margin.
The Expert-Public Gap
Pew’s 2024 expert comparison is the sharpest set of numbers in the whole roundup: 56% of AI experts expect a positive impact on U.S. society over 20 years, against 17% of the general public. Yet the two groups agree on one thing — majorities in both feel they have little or no control over AI in their lives, and majorities in both want more. That points at a product lever: explainability and a visible off-ramp close the trust gap better than raw capability does. Perceptions of whose views count in AI design also split: 40% say White adults’ views are considered at least somewhat well, versus 25% for Asian adults, 19% for Black adults, and 17% for Hispanic adults, with another 37% to 44% unsure.
What It Means for Product Teams
First, the trust deficit is a real design constraint. Half your users arrive worried, and any feature that extends AI’s reach — especially into children’s products or emotional-support contexts — will be scrutinised accordingly. Teams building for teens should treat the 12% emotional-support figure and the six-in-ten cheating perception as hard design lines. Second, news features deserve conservatism: only 9% of adults get news from chatbots at least sometimes and 75% never do — and among those who do, about half at least sometimes see inaccurate news there, with 16% seeing it often. Third, regulatory trust is a divided issue: 44% of Americans trust the U.S. government to regulate AI and 47% do not, but the split runs 54% among Republican leaners versus 36% among Democratic leaners — do not assume one narrative works across the U.S. market. Fourth, age is a more useful segmentation axis than “users” as a blob: ChatGPT use at work among 18-to-29-year-olds (38%) is more than double the 50-plus rate (18%), so products aimed at younger users can assume AI-native defaults, while older audiences need clearer explanation, demonstration, and an exit.
Sources
- Key findings about how Americans view artificial intelligence — Pew Research Center
- How the U.S. public and AI experts view artificial intelligence — Pew Research Center
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.