Two-Thirds of Americans Say AI Is Advancing Too Quickly, Pew Study Finds

ChatGPT usage has more than doubled since 2023, with 44% of U.S. adults now reporting they have used OpenAI's chatbot, according to Pew Research data cited by both The Verge and TechCrunch. Overall chatbot usage jumped from 33% in 2024 to 49% in 2026, marking a dramatic adoption curve even as public sentiment sours.

The heaviest users fall in the 30-to-49 age bracket, where 34% turn to chatbots once daily or more, often for work tasks. About a quarter of all Americans now use AI chatbots daily. Yet nearly half the country, roughly 50%, still reports no AI usage at all, with that abstention concentrated heavily among Americans aged 65 and older. Nearly 75% of seniors say they never use chatbots.

Americans under 30 present a striking paradox: they use AI the most and trust it the least. Pew found that 66% of 18-to-29-year-olds report using chatbots, but 48% believe AI will have a negative impact on society, and just 14% foresee a positive one. This generational reversal upends the usual pattern where younger cohorts embrace new technology most enthusiastically.

Older generations follow the opposite trend: they use AI less but view it less negatively. The data suggests familiarity breeds contempt, at least for now. Younger users encounter AI's limitations firsthand, from hallucinated search results to sycophantic chatbot behavior, and apparently conclude the technology is being deployed faster than it matures. Their daily exposure strips away novelty and surfaces friction points that occasional users miss.

Americans distrust every institution positioned to manage AI's risks. Pew found 67% do not believe the U.S. government will meaningfully regulate AI, while 59% do not trust companies to develop it safely. This dual institutional failure leaves the public feeling exposed, with no credible referee in sight.

The skepticism spans political and demographic lines, suggesting it stems from lived experience rather than partisan framing. TechCrunch notes that Wall Street's AI enthusiasm, exemplified by a hot IPO summer for companies like Anthropic and OpenAI, runs directly counter to this public mood. The gap between investor optimism and citizen wariness now constitutes a genuine political and market risk. Companies building consumer-facing AI products operate in an environment where their own user base doubts their intentions.

Six in ten Americans now routinely read AI-generated internet summaries, Pew reports, a figure that reflects Google's aggressive integration of AI overviews into search results rather than any active user choice. These summaries have become functionally unavoidable. A smaller subset uses AI for fitness and dieting information.

The reliance on AI-curated information carries documented risks. The Verge cataloged multiple instances of AI systems serving inaccurate or fabricated content, from confused dates to dangerous medical advice. Pew's own 2024 study found 66% of U.S. adults were already concerned about AI's role in spreading misinformation. The gap between usage and verification is widening: people consume AI-generated content because it is the path of least resistance, not because they find it reliable. This passive dependency may prove harder to reverse than active tool adoption.

Men use AI more frequently and report greater enthusiasm than women, Pew found. The daily usage gap stands at 27% of men versus 20% of women. While ChatGPT draws equal shares of both genders, men disproportionately adopt competing products: Copilot and Grok show notably male-skewed usage patterns.

ChatGPT dominates the market at 44% penetration, followed by Gemini at 24%, Copilot at 17%, and MetaAI at 14%. Grok, Claude, and Character.ai trail significantly at 8%, 6%, and 3% respectively. The concentration around OpenAI's product suggests first-mover advantage remains potent even as alternatives proliferate. The gender divergence in secondary tool choice hints at different user journeys, with men more likely to experiment beyond the default option. Whether this translates into divergent skill development or economic outcomes remains unmeasured but worth tracking.

The Pew data sketches a society caught in adoption without consent: using AI because it is embedded in work tools and search results, not because it is loved or even fully trusted. The 63% who say AI is advancing too quickly represent a potential political constituency for regulation, even if that same majority doubts government capacity to deliver it.

For AI companies, the implication is stark. User growth no longer signals social license. OpenAI, Google, Microsoft, and Meta now face a population that uses their products daily while believing those products may harm society. That contradiction cannot sustain indefinitely. Either product quality and safety improve visibly enough to close the trust gap, or regulatory and political pressure will intensify beyond the industry's control. The window for self-correction is narrowing as public patience frays.