Half of US Adults Now Use AI Chatbots — But Only 16% Think AI Is Good for Society

They use it but don't trust it — AI's strange coexistence

Here's the deal: Pew Research's 2026 AI survey, released June 17, shows a fascinating contradiction. 49% of US adults use AI chatbots — up 16 points from 33% a year ago. Roughly one in four use them daily. AI chatbots are no longer an early-adopter toy; they're an everyday tool for half the population.

But another number in the same survey flips the mood. Only 16% think AI will have a positive impact on society over the next 20 years. Meanwhile 40% see it as negative. Half use it, but a sixth trust it. A gap this wide between adoption and trust captures AI's strange position perfectly.

Why does the gap exist, who's using it and how, and what does it mean for AI companies and society? Let's unpack it.

What the survey flags — adoption, use cases, and the age gap

First, adoption surge. US adult chatbot use jumped from 33% to 49% in a year. The most used is ChatGPT (44%), then Gemini (24%), Copilot (17%), and Meta AI (14%). ChatGPT still dominates, but Google, Microsoft, and Meta all posting double digits shows a market going multipolar.

Second, use cases. 42% of US adults use AI to research information and 38% for work. AI has moved past "a toy to ask neat questions" into the most practical zones — search and work. It's burrowing into the core seats of legacy IT: the search engine and the productivity tool.

Third, a stark age gap. Among 18–29s, 66% use AI; it drops to 61% for 30–49, 42% for 50–64, and 23% for 65+. Younger means more usage, as expected, but the near-threefold spread stands out. AI adoption isn't an all-generation event; it's a wave the young are pulling forward.

Implications — AI companies, users, and society

For AI companies, this is a double-edged sword. 49% adoption is good news of a fast-growing market. But 16% "good for society" means people use the product without trusting the company or the technology. That trust gap can become regulatory and public-opinion risk. Growing users who are wary of AI is not a pure win.

For users, the gap may be a rational stance. They use it because it's a useful tool, but worry about its effects on jobs, privacy, and the reliability of information. "Using" and "approving" are separate — a pragmatic distance that neither blindly trusts nor rejects the technology.

For society, the survey shows AI growing both convenience and anxiety at once. Adoption is fast but trust lags, so closing the gap falls to transparency, safety, and regulation. The three meet at one point — AI is useful but not yet trusted — and that's the core of this survey.

Echoes of the past — new tech is always "convenience first, trust later"

A gap between "use" and "trust" isn't new. Recall early social media: everyone used Facebook and Twitter while being skeptical it was good for society. Usage exploded but trust came slowly, sometimes even fell. Privacy and misinformation fights eventually drew regulation.

Technologies that closed the gap shared "time and mechanisms to build trust." Online payments scared people at first, but security standards and consumer protection brought trust along. Tech that failed to build trust kept being used but stayed an object of wariness. AI stands at that fork now.

But AI's speed is different. Social media's trust gap surfaced over years; AI's adoption jumps 16 points a year while trust sits still. The gap widens faster and wider. That speed itself is a new variable that regulation and public debate struggle to keep up with.

Competitor counter-play — a second race, over trust

AI companies aren't blind to the trust gap. They front messages of "safety," "responsible AI," and "transparency." Alongside the performance race, a trust race has begun. Some make safety a brand differentiator; some emphasize open source and transparency. That 16% trust figure becomes a battlefield of marketing and regulatory response.

Regulators and civil society are countering too. With adoption past half, voices grow that AI should be treated not as "lab tech" but as near-public infrastructure. Some coverage reads the survey as a sign that public skepticism could turn into regulatory pressure. You can't stop the use; how to manage it is the new question.

For latecomers and new AI services, this may be an opening. If trust in incumbent giants is low, there's room to differentiate as "more transparent, safer AI." Performance alone won't cut it; baking trust into design may pull new users.

So what actually changes

If you build AI services, this is a clear signal. Users grow but trust doesn't, so weaving transparency, safety, and control into the product matters more and more. Beyond "AI that works," "AI you can trust" becomes the differentiator.

If you use AI, the gap may be a healthy sign. Using a tool without blind faith is rational. When using it for research or work, building a habit of verifying its answers is the safe move.

If you watch policy and society, note that the gap between 49% adoption and 16% trust will be a key variable for regulation and opinion. How that gap is closed will likely shape how AI settles into society.

One step further — what "use it but don't trust it" really signals

What the gap between 49% and 16% really says is that AI is a rare technology growing "need" and "distrust" at once. Usually trust accumulates first, then usage follows. With AI the order is reversed: before being fully trusted, half started using it simply because it's useful. That's because AI has such a low barrier to entry and gives immediate utility — tossing one question at ChatGPT needs no trust. So a state rare in past tech emerged: mass use without trust.

Why it matters: this gap is the AI industry's biggest latent risk. Even with many users, if they're wary of the technology, a single small incident can flip opinion sharply. If misinformation, privacy violations, or job shocks blow up once, that 40% negative sentiment can erupt into regulatory pressure. Social media walked exactly that path — everyone used it while distrusting it, and when something like Cambridge Analytica hit, regulation and backlash piled in at once. AI now carries the same powder keg.

But this distrust shouldn't be read as purely negative. "Use it but don't blindly trust it" can actually be a sign of healthy digital literacy. If people know AI can produce wrong answers (hallucinations) and stay wary, that's the maturity of consuming technology critically. The question is whether that wariness leads to rational verification or hardens into vague rejection. The former grows a healthier AI ecosystem; the latter can trip up even useful technology with distrust.

There's also the social split the age gap creates. The threefold gap — 66% for 18–29s, 23% for 65+ — shows AI could become a new dividing line in cross-generational information and work skills. While the young search and work with AI, if older people drift from that flow, the "digital divide" expands into an "AI divide." That's not mere taste; it can become real inequality in jobs, information access, and public services. So AI adoption is hard to read as pure progress.

Finally, the implication for AI companies is clear. Performance alone won't lift that 16%. Whoever weaves transparency (how answers are made), safety (preventing wrong and harmful answers), and control (users managing data and outputs) into the product takes the trust. The next AI battle's stage likely shifts from benchmark scores to "how trustworthy is it." This survey shows that turning point in numbers.

🥄 Three Things You're Probably Wondering

— So what does this mean for me? If you already use AI chatbots, you're in the half. Just remember "using" differs from "trusting," and building a habit of verifying answers lets you use it more safely.

— Why use it but not trust it? Because usefulness and anxiety are separate. It's handy as a tool, but you worry about its effects on jobs, privacy, and information reliability. That's not irrational — more like pragmatic distance.

— Will the trust gap close? Too early to say. Past tech closed gaps with security, regulation, and time, but AI's adoption is so fast that trust struggles to catch up. It depends on how quickly transparency and safeguards take hold.

References

• Americans and AI 2026: Chatbots, Smart Devices and Views on Impact — Pew Research
• How opinions and use of AI differ by age — Pew Research
• Just 16% of Americans think AI will benefit society, despite chatbot use climbing to 49% — TechSpot
• Just 16% of Americans Believe AI Will Positively Impact Society, Pew Poll Finds — Gizmodo
• Pew Study Finds Americans Fear AI's Societal Impact — Variety

Numbers and criteria are as of announcement and may change.