AI has a reputation problem in the US. In Pew's June 2026 survey, 52% of Americans were more concerned than excited about its growing use in daily life. Only 9% felt the reverse. (Pew Research Center)

The complaints are familiar: training on other people's work without permission, energy use, an internet filling with slop, cheating, concentrated power, and the prospect of entire careers disappearing.

Dario Amodei recently argued that winning public trust requires delivering the promised benefits, including actually curing cancer. He also acknowledged that distrust of companies and governments long predates AI. (Business Insider)

Personally, I've noticed that AI often seems to amplify the outlook people already had. The optimistic people I know, who feel in charge of their lives, are finding more things to build and explore than they have time for. They're going deeper into technical subjects and attempting creative work they previously felt unequipped to do. They're overwhelmed, but mostly by the number of things they want to try.

Among the people I know who have a less optimistic view of the world, the same technology brings familiar worries closer: less secure careers, environmental damage, and more power in the hands of people they don't trust.

I believe part of this is perceived agency: the sense that your choices and effort can meaningfully shape your future. A more capable tool is exciting when you can envision what you could do with it. It is a different proposition when you expect to have little control over how it changes your life.

Research on related concepts suggests a connection. A 2025 study of 562 adults in the UK and Gulf Arab countries measured general self-efficacy: confidence in handling everyday challenges. People reporting greater self-efficacy also reported more positive and less negative attitudes toward AI. (Springer Nature Link)

A separate 2023 paper followed Finnish participants over time and found that increases in perceived autonomy when using technology were associated with more favorable views of AI. (Self Determination Theory)

The associations were modest, and neither study is from the US. However, the Finnish data further found that people whose income rose over the year grew less negative about AI, which fits the idea that circumstances shape how people view it.

Peter Thiel made an argument in a 2020 email about millennials' attitudes toward capitalism. Heavy student debt and unaffordable housing, he argued, left many people unable to accumulate capital of their own, and with little stake in the system, they had less reason to support it. (Business Insider)

I suspect some of the pessimism about AI follows the same logic, and it isn't limited to millennials. For someone facing expensive housing, healthcare and college without much wealth of their own, it's reasonable to wonder whether the next wave of AI-fueled growth will benefit them. Pew found that 71% of Americans expect AI to mean fewer jobs over the next 20 years, and just 5% expect more. (Pew Research Center) Plenty of people who understand what AI can do may not expect it to benefit them personally, or feel they have the room to take advantage of it.

That negative perception is unfortunate, because AI makes creating things cheaper and more accessible. A teenager with a laptop can use the same models as a Fortune 500 company. Similarly, startups have long been disruptive to incumbents because they were small and agile, which made up for having fewer resources. Today, with AI, that resource gap is itself evaporating.

Amodei is correct that delivering tangible benefits from AI is important, and curing cancer is a very worthy goal. But even a cure discovered today, with or without AI, might be met with a mixed reception, as some would wonder if they would ever have access to it.

So the best way to improve how people feel about AI may be to work on the problems that existed long before it: housing, healthcare, and the cost of college. AI may help solve those problems, but all of them are clearly solvable without it. With a plausible path to a home, manageable debt, and some financial security, people have reason to expect their effort to pay off, and more reason to be invested in a system they feel part of.

People who believe they can build a better life have a reason to welcome tools that let them do more. Giving more people grounds for that belief may matter more than finding a better way to sell them on AI.