The Worry and the Reality

About two-thirds of Americans say artificial intelligence (AI) is developing too quickly. At the same time, the number of people actually using AI chatbots like ChatGPT has roughly doubled in just two years. That contradiction is the focus of a new report from Pew Research Center, released in June 2026.

This worry crosses all age groups. People in their 20s, 40s, 60s, and beyond all express similar concerns about the speed of AI development. There is one notable difference: women (68%) are more likely than men (58%) to say AI is moving too fast — a ten-point gap that shows up consistently. This difference may reflect that women encounter AI in certain places more often, such as content filtering systems, hiring tools, and surveillance technologies, which tend to receive more critical attention.

Alongside worries about speed, most Americans also say they're concerned about AI putting their personal information at risk. This combination — worries about both speed and privacy — has appeared before with other big tech shifts. When search engines started personalizing results, when social media collected data from users, and when phones began tracking location, people had similar paired concerns.

But Usage Numbers Are Growing Fast

Yet actual usage tells a different story. In 2025, about one-third of all American adults had tried ChatGPT — roughly double the number from 2023. For people under 30, the number is even higher: 58%. By mid-2026, around 63% of Americans under 50 have used some kind of AI chatbot, suggesting the trend is continuing to accelerate.

Teenagers are particularly notable. By early 2025, 26% of teens aged 13 to 17 had used ChatGPT for homework — up from 13% just two years earlier. That is a very fast adoption rate. To put it in perspective, it took longer for tablets to become common in classrooms, and it took longer for search engines to become normal homework tools. Schools across the country are still deciding how to handle AI in education, with some banning it entirely and others building it into their curriculum.

AI is also expanding beyond chatbots. More people are using AI features that summarize information and smart speakers that respond to voice commands. Often, people don't even realize they're using AI — it's built into search results, email drafts, and voice assistants in the background.

The Paradox Explained

The gap between what people worry about and what they actually use is not really a contradiction. People routinely use technologies they're uncertain about if those tools seem useful for their own purposes — even if they worry about how the company behind the tool operates. Social media is the clearest example: millions of people use Facebook and TikTok despite concerns about privacy and data practices. Another example is ride-sharing apps like Uber and Lyft, which many people use while still being uncomfortable with labor and privacy questions those services raise.

What makes AI different right now is how fast the entire cycle is happening. ChatGPT came out in late 2022. By mid-2025, about one-third of Americans had used it. For comparison, the iPhone was released in 2007, but it took about six years for smartphones to reach the same level of household adoption. The internet took even longer to reach that many people. AI is spreading faster.

When Americans say AI is moving too fast, they probably don't mean they want it to stop entirely — the number of people using it shows that is not what they think. Instead, they likely mean that schools, Congress, and other institutions are not keeping up. Schools are still writing policies on how to handle AI. The U.S. Congress has not passed major AI laws. Courts are still figuring out who is responsible if an AI system makes an error.

For companies building AI products and for people working in government and policy, the takeaway is clear: people like using AI tools and will keep adopting them. But people do not feel confident that the systems managing and overseeing AI are working well. Closing that gap between how much people use AI and how much they trust it — that is the challenge for the next few years.