The AI Regulation Poll Is a Warning About Trust, Not Just Technology
A new Penn Annenberg Public Policy Center survey, reported by The Philadelphia Inquirer on May 16, found that 71% of U.S. adults believe the federal government has not done enough to regulate artificial intelligence. Fewer than one in five respondents said AI would have a positive effect on the United States over the next decade, even as 57% expressed optimism about AI’s role in medical research. The survey of 1,330 adults points to a split that matters politically: Americans are not simply anti-technology. They can see potential benefits in science and medicine, but they do not trust the current institutional bargain among Washington, Big Tech, and the companies racing to deploy increasingly powerful systems. That makes AI regulation less of a niche technology issue and more of a public-trust fight over who captures the upside and who absorbs the risk.
The important part of the Penn Annenberg survey is not that Americans are “worried about AI.” That line is too easy. The more useful signal is that the public can hold two ideas at once. People can believe AI may help medical research and still believe Washington has not done enough to regulate the industry.
That is not contradiction. That is actually a fairly sophisticated read of the situation.
The country has seen this movie before. A new technology arrives wrapped in promises of abundance, productivity, personalization, and inevitability. The firms building it ask for trust while racing for market share. Policymakers hold hearings, discover the vocabulary six months late, and then split into predictable camps: one side wants to posture as pro-innovation, the other wants to posture as pro-safety, and both sides quietly understand that the money and expertise live inside the same companies asking to be regulated gently.
Ordinary people may not follow model weights, chip supply chains, or frontier-lab governance debates. But they understand the basic incentive problem. If the winners of AI are allowed to privatize the gains while socializing the disruption, then this is not a technology story. It is a power story.
That is why the medical-research number matters. A majority being optimistic about AI in medicine shows the public is not rejecting progress. People want better diagnostics, faster drug discovery, less paperwork, and more useful tools for doctors. What they do not want is another platform era where the sales pitch is human flourishing and the business model becomes dependency, surveillance, labor displacement, and institutional capture.
The political class keeps trying to frame AI as a race: race China, race competitors, race the next model release, race the next investor call. There is truth in that frame, but it is incomplete. Speed is not the same thing as legitimacy. A system can move quickly and still lose public trust if citizens conclude that nobody in authority is protecting the downside.
The second-order consequence is that AI policy may stop being a specialist issue. If people begin connecting AI to jobs, schools, health systems, scams, insurance decisions, policing, media manipulation, and energy bills, then the old “leave it to the experts” posture breaks down. The expert class has not earned enough trust to demand a blank check.
The answer is not panic regulation written by politicians who barely understand the tools. It is also not letting the largest platforms write their own rules in the language of safety. The answer starts with admitting that incentives are the core issue: who audits the systems, who pays when they fail, who owns the data, who benefits from deployment, and who gets stuck living under automated decisions they cannot appeal.
That is the public mood this poll is catching. Americans are not asking Washington to stop the future. They are asking why the people promising the future always seem to get rich before anyone else gets accountability.