Two-Thirds of Americans Say AI Is Moving Too Fast, Even as Chatbot Adoption Doubles

Roughly two-thirds of Americans believe AI is advancing too quickly — even as the same population is adopting the technology at a pace that has roughly doubled in two years. That is the central tension in Pew Research Center's "Americans and AI 2026" report, published June 17, 2026.

The unease is not confined to any particular demographic. Pew's age-segmented analysis found that majorities across all adult age brackets share the sentiment, with the concern registering at 61% or higher regardless of generation. The gender split is sharper: 68% of women versus 58% of men say the pace is too fast — a ten-percentage-point gap that has shown up consistently in Pew's AI tracking and likely reflects differing exposure to AI-driven content moderation, hiring tools, and surveillance applications that have attracted the most critical coverage.

Alongside the pace concern, majorities also told Pew they believe AI will put their personal information at risk. That pairing — speed anxiety and data risk — is consistent with how public sentiment has formed around prior platform-scale shifts. Search engine personalization, social media data collection, and smartphone location tracking each generated similar dual concerns before regulatory or market pressure forced at least partial resolution.

Adoption Numbers Tell a Different Story

Against that backdrop of unease, actual usage has climbed steeply. A 2025 Pew survey found 34% of US adults had used ChatGPT — approximately double the figure recorded in 2023. Among adults under 30, that number jumps to 58%. By mid-2026, Pew data puts chatbot usage among Americans under 50 at around 63%, suggesting the adoption curve has continued accelerating into this year.

The teen cohort is also notable. By early 2025, 26% of US teens aged 13 to 17 reported using ChatGPT for schoolwork — up from 13% in 2023, a doubling in roughly one academic year. For context, that rate of educator-relevant adoption outpaces what was observed in the early years of both tablet deployment in classrooms and consumer search engines finding their way into homework. The K-12 implications are still being worked out institutionally, with policies ranging from full bans to formal integration varying by district.

Pew also recorded broader growth in AI summary features and smart speaker usage, consistent with the expansion of AI beyond the explicit "chatbot session" into ambient and inline interfaces — search results, voice assistants, email drafts — where users may not consciously register the interaction as AI at all.

The Gap Between Concern and Behavior

The divergence between stated anxiety and actual usage is not a contradiction so much as a familiar feature of platform adoption cycles. People routinely use technologies they distrust at the institutional level while finding individual utility compelling enough to override that distrust. Social media is the most documented recent example; ride-hailing and its privacy and labor implications is another.

What makes the current AI moment distinct is the compression of the cycle. The gap between ChatGPT's public release in late 2022 and a third of the adult US population having used it was roughly two and a half years. For comparison, it took smartphones closer to six years to reach equivalent household penetration after the iPhone's 2007 launch, and broadband internet took longer still. The infrastructure was different, but the diffusion rate for AI-native applications is plainly faster.

That speed is precisely what the Pew data captures in the public's own framing. When nearly two-thirds of Americans say AI is advancing too quickly, they are not necessarily calling for a halt — the usage numbers make that reading implausible. They are more likely registering that the social, regulatory, and institutional scaffolding has not kept up. Schools are still forming chatbot policies. Congress has not passed comprehensive AI legislation. Liability frameworks for AI-generated outputs remain unsettled.

For practitioners building products, deploying models in enterprise workflows, or advising on AI governance, the practical read of this survey is straightforward: public tolerance for AI integration is high enough to sustain continued adoption, but trust in the broader system governing AI is not. That gap — between product-level utility and institutional-level confidence — is where the next several years of policy and technical work will concentrate.