How Americans view AI in 2026

A look into a new Pew study

Greetings from a very hot mid-Atlantic summer. Temperatures and humidity have steeply risen, causing plans to flourish and humans to languish. We’re closing in on the heat which has stricken western Europe as I write this.

Today I’d like to share an important new survey concerning AI. It’s from the excellent Pew Research, a resource I’ve relied on for decades, and focuses on American use of and attitudes towards the technology and its implications. I found it striking but also resistant to a summary analysis, so rather than folding it into a scanner report I’m devoting an entire post to it.

I’ll break down what I see as the highlights, divided into categories of AI use and attitudes, followed by demographic details, then offer some future-oriented reflections.

NB: this post will have a lot of graphics, more than my usual, because Pew does a great job with its charts.

tl;dr version - more Americans are using AI than before, and in many different ways, while at the same time most Americans are critical of the tech.

Quick details: Pew surveyed people in February 2026, so things may have shifted since then. Their sample size was 5,119. Here’s a link to the study’s topline.

1 AI use and attitudes

Let’s start with the basics of AI use. About half of Americans use AI; about half of those use it daily. Another half of the nation says they don’t use AI.

Pew asked the non-users why they avoid the tech, and I was fascinated by the responses. Most simply had no interest in AI. Nearly as many were concerned about their privacy. And a lot of people suspect the information quality:

Note how relative few people say they don’t know how to use the tech. And how a small slice worries about being judged.

What do we use AI for, or at least what are we willing to tell Pew we do with the technology? Information search continues to lead the pack, followed by work use cases:

For that leading use case, more than half of Americans describe using AI search summaries:

(I’ll note that 10% unsure probably includes some people who do so but aren’t aware of it. Perhaps 2/3rds of us should be the estimate.)

In terms of individual AI tools ChatGPT remains the most popular chatbot, with Gemini in second place, CoPilot in third:

While a good number of Americans are using AI, a preponderance of us express negative sentiments about the tech. There is no majority view in Pew’s report, but more folks are critical than positive:

Note the big chunks who see AI as equally good and bad, plus those unsure.

On a related note, Americans were even more skeptical about AI’s pace of development. “Americans largely think AI is moving too fast. About two-thirds say this.Only 2% say it’s advancing too slowly.”

That’s a solid majority of opposition, at least in terms of speed. In contrast, accelerationists are vanishingly rare.

There’s a similar solid bloc who are worried about AI and security: “roughly seven-in-ten predict AI will make their personal information less secure. Just 3% say it will lead to their information being more secure.”

Americans generally distrust those in charge to handle AI well. Both governments and corporations are under suspicion on this score:

67% of Americans have little to no confidence in the U.S. government to regulate AI effectively. This is slightly up from 62% in 2024, when we last asked this question.

When it comes to trusting the businesses who develop AI, about six-in-ten adults are not confident in U.S. companies to develop and use these tools responsibly. [emphases in original]

Political party affiliation showed some predictable attitudes, with Democrats more skeptical about government AI work now that Republicans are in charge:

2 Demographic breakdowns

Pew broke down its population by demographics, which showed interesting differences. Segmented by age, we see a traditional tech spectrum in terms of usage, with younger people more likely to use the stuff than their elders, a pattern Pew has noted for two years:

Of that population, 32% of the youngest group use AI daily, which is higher than the total population number. For those of us tracking companionbots, this was an important note: “one-in-five adults under 30 say they ever turn to chatbots for emotional support or advice.” (Think of the implications for teaching traditional-age undergraduates.)

But attitudes ran the other way by age, with younger people more likely to be critical of AI than older folks:

Turning the demographic analysis to gender, Pew offers further findings. For one, while there was a gender gap over AI use during the past few years (men more likely to use AI than women), that divide has now closed:

Yet women and men tend to express different attitudes towards the technology, with women being more likely to see AI as negative for themselves and society:

Pew hit this point strongly in their report:

Across our surveys, there’s a consistent pattern where women tend to view AI more negatively than men. These gender differences also show up in other surveys of teenagers and AI experts. [emphases in original]

Other gender differences persist. Pew also found more women than men viewing AI’s development as too fast. Daily use patterns show a divide, as “[m]en are more likely than women to say they use chatbots on a daily basis (27% vs. 20%).” Attitudes towards AI and productivity also diverge, since “[w]hile 35% of men say these tools help their productivity, the share drops to 25% among women.” There’s a slight split over AI in search as well: “Larger shares of men report reading AI summaries that appear at the top of search engine results, compared to women (63% vs. 57%).” And another small split over the one area where women are more likely to use AI than men, for “emotional support or advice,” 8% of men, 11% of women.

Additionally, Pew found an interesting gender divergence by specific AI tools: “men are more likely than women to say they ever use Copilot (22% vs. 13%). Larger shares of men than women also report using Gemini, Grok and Claude.” In contrast “ChatGPT… is used by identical shares of men and women.”

Turning to race, Pew found Asian-Americans using AI far more often on a daily basis than people of all other racial groups:

That daily use is about double! Moreover, Asian respondents used AI for more purposes than other people, overall:

I was struck by news, diet, and medical functions, where a significant proportion of Asian-Americans outnumber other populations quite starkly.

Asian-Americans also view the technology more positively than everyone else. They are, as far as I can tell, the only racial group with net positive attitudes to AI:

Pew also included economic differences in some answers. Importantly, wealthier people tend to be more positive about AI than poorer folks:

(annotation is mine)

Chatbot use is similarly marked by class, with wealthier people far more likely to say they’ve fired up the tech.

---

3 Reflections

So what does this survey tell us about the future of AI, with an eye on the academy?

Once again America, the world leader in making AI, has a lot of us using the technology, while also being strongly critical of it. Perhaps we’re seeing pro- and anti-AI attitudes and practices deepening and hardening, a divide to expect in years to come. Divides by gender, class, age, and political party are appearing.

I was struck by how many people who don’t use AI aren’t interested in changing their minds:

Most who don’t use chatbots say they won’t use these tools in the near future.

Roughly seven-in-ten adults who don’t use chatbots say they’re unlikely to use the technology in the next 12 months. This includes 40% who say they’re not at all likely to.

Only 5% of chatbot non-users think it’s highly likely they’ll use a chatbot over the next year. Another 16% say they’re somewhat likely. [emphases in original]

America is increasingly split over AI, in other words, and appears likely to stay that way. At the same time we should heed the substantial numbers of people who are unsure, undecided, or seeing the tech as a wash. There’s room for America to break decisively in one direction or another.

In terms of use, it seems important that information-finding and work are the leading functions performed, more so than creating and editing content or conversation. AI’s replacing web search (i.e., Google) is proceeding apace, with all kinds of implications: pressures on web content owners and creators; new forms for information literacy; AI as a rising source for new.

In terms of market share OpenAI is still in the lead, or was in February, when Pew did this study. I am curious to see if Anthropic enjoyed a big user expansion in the course of its struggle with the American government.

Those numbers of skeptics and especially the solid majority opposed to AI’s buildout speed suggest to me some room for political positioning. We might expect local and state officials and candidates to float slowdown options. Colleges and universities, already often slow to grapple with the technology, have many ways to slow AI’s roll. The widespread concern about AI’s impact on security might signal a population willing to take or put up with more safety measures. Negative attitudes towards government work on AI open up opportunities for politicians to take strong oppositional stands. Similar attitudes to corporations leave openings for some companies to break for the herd (viz Anthropic) as well as for open source.

Kudos to Pew for continuing to conduct this vital research.

(thanks to John O’Brien for the nudge)