63% of Americans Fear AI Is Racing Ahead as Chatbots Spread

What does it say about the AI boom that 49 percent of Americans use chatbots at least occasionally, yet 63 percent think AI is advancing too quickly?

That tension is the story. The latest Pew Research Center poll, reported by The Verge, shows a public that is using AI chatbots more often while becoming harder to persuade that the rollout is healthy, controlled, or socially beneficial.

XOOMAR analysis: AI sellers get squeezed when the industry treats adoption as proof of acceptance. Pew’s numbers cut against that. Curiosity, convenience, and task-level experimentation don’t equal trust. Americans are trying the tools, but many don’t like the pace or direction of the buildout.

Why do Americans think AI is advancing too quickly while still using it?

The obvious answer is that usefulness and confidence are different things.

Pew found that 49 percent of Americans report using chatbots at least occasionally, up from 33 percent in 2024. That’s a sharp rise over two years. ChatGPT has become the clearest brand beneficiary: 44 percent of respondents said they’ve used it, roughly double the share reported in 2023.

But the same polling shows deep discomfort:

That 16 percent optimism figure is brutal for a technology still being sold with maximalist language. A tool can spread because it is available, embedded, free to try, or useful for a narrow job. That doesn’t mean people approve of the broader AI buildout.

Occasional use also deserves precision. Someone asking a chatbot to draft an email once a month is not the same as someone relying on it daily for work. Pew’s adoption curve is real, but it doesn’t prove dependency. It proves exposure.

For readers evaluating AI by task rather than hype, that distinction matters. Narrow uses, such as the spreadsheet work covered in ChatGPT vs Claude Spreadsheets Test Picks Clear Winners or ChatGPT’s science lab plan, are a different question from whether the public trusts AI’s role in society.

Why are young Americans heavier AI users and sharper AI skeptics?

The most revealing split is generational.

Pew found that 66 percent of Americans ages 18 to 29 report using chatbots. That makes younger adults the highest-use age group. Yet 48 percent of that same group believe AI will have a negative impact, while only 14 percent believe it will have a positive one.

Older Americans report using chatbots less. They also hold less negative views, according to the Verge summary of Pew’s findings.

That undercuts the lazy assumption that younger users are natural AI evangelists. Familiarity can breed criticism. People who touch the tools more often may also see the errors, limits, and social friction more clearly.

Pew also found that daily use is highest among Americans ages 30 to 49, with 34 percent saying they use chatbots once a day or more. One possible driver is work: roughly four in ten Americans reported using AI for work tasks. 30 percent said AI makes them more productive, and 28 percent said it helps them be more informed.

But the upside sits next to a credibility problem. In Pew’s 2024 study, 66 percent of US adults said they were concerned about AI spreading inaccurate information. That concern fits the current mood: people may use AI because it helps with a task, while still doubting the output and the incentives behind it.

Is this mainly a Pew outlier, or does other polling show the same trust gap?

Other surveys point in the same direction, though the numbers differ.

A separate Economist / YouGov Poll found that 71 percent of Americans say the pace of AI development is moving too fast. 27 percent said the pace is about right, and only 2 percent said it is moving too slowly.

That poll also found economic skepticism. 64 percent of Americans said it is unlikely AI will create economic gains that benefit everyone. Only 8 percent said that outcome is very likely, while 28 percent said it is somewhat likely.

The job anxiety is not evenly distributed:

• Young adults: 60 percent of adults under 30 are somewhat or very worried that AI will replace jobs they rely on.
• Lower-income households: 56 percent of Americans with family incomes under $50,000 are somewhat or very worried about AI replacing jobs they rely on.
• Broader sentiment: 51 percent of Americans are pessimistic about AI’s long-term impact on society, compared with 25 percent who are optimistic.

That helps explain why Americans can be active users and still say AI is advancing too quickly. They are not only judging the product in front of them. They are judging who benefits, who bears risk, and whether institutions can keep up.

“Americans are not rejecting AI outright, but they are sending a warning,” Associate Professor of Business Analytics and Information Systems Tamilla Triantoro wrote. “Too much uncertainty, too little trust, too little regulation, and too much fear about jobs.”

That quote captures the core distinction. The public isn’t voting to unplug AI. It is signaling that adoption without accountability has a ceiling.

Which AI risks matter most to workers, companies, and regulators?

The risk map changes depending on who is looking.

For workers, the YouGov data points to job exposure and economic distribution. People are not only asking whether AI can make tasks faster. They are asking whether the gains will flow broadly or concentrate elsewhere. The correlation in YouGov’s findings is telling: among Americans who say it is unlikely AI will economically benefit everyone, 69 percent feel pessimistic about AI’s long-term effects.

For companies, Pew’s productivity numbers are tempting. If roughly four in ten Americans already use AI for work tasks, and 30 percent say it makes them more productive, firms have a reason to keep testing tools. But forced or poorly explained AI adoption risks running into employee skepticism, customer distrust, or both.

For regulators, the trust deficit is already visible. CNET’s summary of Pew’s findings noted that 67 percent of US adults have little to no confidence in the US government to effectively regulate AI, while about 60 percent are not confident US companies will develop and use AI tools responsibly.

That is the hardest combination for the industry: low trust in government and low trust in companies.

Task-specific AI will still have room to grow. A proposal team may weigh risks differently than a customer service department or a solo consultant. That’s why practical evaluations such as AI Grant Writing Tools Can Save Proposals or Sink Them and AI Productivity Apps for Consultants That Cut Busywork are more useful than broad declarations about whether AI is “good” or “bad.”

Can AI companies turn usage into legitimacy?

They can, but Pew’s numbers show they have not done it yet.

XOOMAR analysis: The next phase of AI competition won’t be defined only by model capability. It will be shaped by whether products feel optional, understandable, and accountable. Users need to know when AI is present, what data it touches, and where human review still matters.

The public opinion pressure can affect product design, workplace policy, school rules, labor negotiations, and regulatory agendas. That doesn’t require everyone to reject AI. It only requires enough people to distrust the rollout.

The evidence that would confirm this thesis is straightforward: continued growth in chatbot use alongside stagnant or worsening optimism, more concern about job replacement, and persistent doubts about companies or government oversight.

The evidence that would weaken it would be just as clear: rising confidence that AI benefits are broadly shared, stronger belief that companies use AI responsibly, and a higher share of Americans saying AI will have a positive impact on society.

For now, adoption is rising faster than legitimacy. That gap is the problem AI builders can’t solve with another product demo.

The Bottom Line

• AI adoption is rising, but usage does not equal public trust.
• The gap between chatbot use and optimism suggests tech companies face a credibility problem.
• Public concern about AI’s pace could shape future regulation, product design, and adoption.