A new Pew Research Center survey of 5,119 US adults finds 49% now use AI chatbots while 40% expect the technology to make society worse, an inversion of how the internet, the smartphone, and social media all spread.
According to a Pew Research Center survey of 5,119 US adults conducted in February 2026, 49% of Americans now use AI chatbots, up from roughly a third in 2024 based on Pew's prior tracking data. That same survey found 40% expect the technology to make society worse, 63% say it is advancing too fast, and majorities have little confidence in either the companies building it or the government supposed to regulate it. The numbers describe an inversion with no clean historical precedent in American consumer technology.
AI chatbots are text-based assistants such as ChatGPT, Microsoft's Copilot, and Google's Gemini, software that can answer questions, draft documents, write code, and carry on a conversation. Their adoption has been the steepest in Pew's tracking of any digital tool since the smartphone. Unlike every comparable rollout, however, the curve has been climbing against public skepticism rather than enthusiasm.
The historical record offers only rough parallels, and they come with a measurement caveat: the relevant benchmarks — household internet penetration crossing 50% in the early 2000s, US smartphone ownership reaching roughly half of adults in the early 2010s, social media adoption crossing the same threshold earlier in that decade — use different units than an individual adult chatbot-use rate. The dominant public mood during each of those adoption curves was optimism: a tool that would connect the world, spread knowledge, and democratize commerce. Social media, which crossed that threshold earlier in the same decade, is now the cautionary tale that AI was supposed to avoid, not the model it would replicate.
The June 2026 release suggests that has not happened. Of those who expect AI to make society worse, the largest share cite job loss, surveillance, and the spread of misinformation. The skepticism is not uniform. Younger, more educated, and higher-income respondents are both more likely to use chatbots and more likely to say AI will make things better. Concern, however, cuts across age, education, and party, with majorities of Democrats and Republicans alike saying AI is moving too fast.
Trust in the institutions supposed to manage the technology is even lower than trust in the technology itself. The survey found 59% of US adults have little or no confidence in the companies developing AI to do so responsibly, and 67% have little or no confidence in the federal government to regulate it. That puts the burden of trust on a third party the survey did not directly ask about: the employers, schools, and product teams that decide what AI actually does in daily life.
Pew's tracking shows the concern has been building for years. The center's April 2025 report on expert and public views of AI found a wider gap between AI researchers and the general public than on any prior technology, with experts broadly optimistic and the public broadly worried. A September 2025 follow-up found pessimism hardening. A March 2026 short read reported a growing share of Americans saying AI would worsen their own lives specifically, not just society in the abstract.
The June 2026 data appears to mark a first in Pew's surveys: adoption of AI chatbots and distrust of the technology have both crossed the 50% mark simultaneously. That makes the trust gap a structural break in how a major general-purpose technology gets adopted in the United States. The institutions now under the most pressure are not the model builders, which the public already discounts, but the employers, schools, and product teams that decide what AI actually does in workplaces and classrooms. Whether that gap narrows, widens, or simply gets normalized into the background of American life is the question Pew will be asking this time next year.