AI Was Meant to Win Us Over. New Data Shows It's Doing the Opposite.

Key points

• A Pew Research study published this week found 52% of Americans say they are "more concerned than excited" about AI in daily life, up from 37% in 2021.
• A May 2025 Economist/YouGov poll found more than 70% of Americans believe AI is advancing too fast.
• Anthropic CEO Dario Amodei publicly called negative public perception of AI a "crisis of trust" in a post on X this week.
• Tech companies building AI data centers across the US are now offering local communities cash bonuses, job guarantees, and clean water investments to reduce opposition.
• Airbnb CEO Brian Chesky said the AI industry needs to build products that "regular people" actually want, not just tools that impress insiders.

The numbers were supposed to go the other way. The more AI spread into everyday products, the thinking went, the more people would come to appreciate it. That is not what is happening.

Pew Research published a study this week showing that 52% of Americans describe themselves as "more concerned than excited" about AI becoming a bigger part of daily life. In 2021, that figure was 37%. The direction of travel is clear, and it is away from enthusiasm.

A separate Economist/YouGov poll from May found that more than 70% of Americans think AI is moving too quickly. A recent CNBC survey of adults aged 18 to 34, a group the industry had counted on as natural early adopters, found that a majority do not trust the top names in AI leadership to act responsibly.

Why are people so uneasy?

Most people are not thinking about AI as an abstract research project. They see it as the chatbot that summarises their emails, the search result that replaced the actual web page, or the tool their children use to write essays at school.

The trade-off feels lopsided. AI features have arrived inside televisions, inboxes, and search engines, whether users asked for them or not. Meanwhile, the promised upsides, fewer working hours, higher pay, cures for disease, have not materialised for most ordinary people. What has materialised is anxiety about job security, questions about the value of a university degree, and a sense that vast amounts of creative work, from writing to music to art, is being fed into AI systems without the original creators' permission.

That frustration is showing up in unexpected places. Sales of so-called dumbphones (basic handsets with no internet browser) are climbing. Classic iPods with no algorithm are fetching high prices on resale sites. Run clubs, knitting circles, and in-person card games are filling up. These are not coincidences. They look like deliberate choices by people who want less technology, not more.

What does this mean in practice?

It is becoming a financial problem, not just a feelings problem. Tech companies are facing local opposition to the large data centres they need to run AI systems, and they are now paying to smooth things over. The Wall Street Journal reported this week that deals now include job guarantees, investments in local water infrastructure, and, in one case in Louisiana, $50,000 bonuses for teachers in a single parish.

Even at the top of the industry, the mood is shifting. Airbnb CEO Brian Chesky said recently on a podcast that the backlash is real, and that it comes down to the industry failing to build things ordinary people find useful. "We need more regular things," he said, giving the example of an on-demand doctor that someone who cannot afford private healthcare could actually use.

Dario Amodei, CEO of Anthropic, the AI safety company behind the Claude family of AI assistants, was more direct. In a post on X this week, as reported by TechCrunch AI, he called the situation a "crisis of trust" and said the most accurate criticism of AI companies is simple: "We haven't yet delivered on our big promises to benefit the world. That is totally on us."

For patients and anyone hoping AI will genuinely improve healthcare, that admission matters. The tools that could make a real difference, better cancer screening, faster drug discovery, earlier diagnosis, are still largely in research settings. What has reached the public first is the easy stuff: chatbots and content summaries. The hard, genuinely useful work is taking longer.

Common questions

Is AI actually dangerous, or is this just hype?

Most current concerns are not about science-fiction scenarios. People are worried about practical things: job losses, AI-generated misinformation, children's education, and companies collecting data to train systems. Those are real questions worth asking.

Should I be worried about AI in the products I already use?

You do not need to panic, but it is reasonable to check the privacy settings on apps that have recently added AI features, particularly around whether your data is used for training. Most products now include an opt-out option somewhere in settings.