Two in every three Americans think the government has done too little to rein in artificial intelligence. That figure is not, in itself, surprising. What is surprising, in this particular political moment, is who those Americans are: Democrats, independents, and Republicans, in roughly equal measure. On almost every other issue you could name, the partisan gap runs somewhere between 65 and 76 percentage points. On AI regulation, it shrinks to almost nothing.
This is the central finding of a new nationally representative survey from the Annenberg Public Policy Center at the University of Pennsylvania, conducted among 1,330 adult citizens this past February and March. The researchers were asking about the future. And the future, according to most Americans, looks fairly grim.
A Technology People Use but Don’t Trust
Nearly 8 in 10 respondents said they had heard at least a moderate amount about AI. About two-thirds reported using it, in some form, at least a few times in the past month. These are not people who are strangers to the technology. And yet when asked what they thought AI’s overall impact on the United States would be over the next decade, only 17% said positive. Forty-two percent said negative. Thirty-two percent, perhaps the most interesting group, said equally positive and negative, which is a very particular kind of ambivalence: not ignorance, not indifference, but a sort of eyes-open dread of something that might cut both ways.
The one domain where genuine optimism broke through was medicine. Over half of those surveyed expected AI to have a positive impact on medical research and discoveries, which makes a sort of instinctive sense. AI-assisted drug discovery, faster diagnostic imaging, the promise of genomic pattern recognition at a scale no human team could manage: these are the stories that have filtered through. But drop down from medicine to almost anything else and the numbers fall sharply. Only 22% expected AI to benefit the creative arts. Nineteen percent thought the economy would improve. For mental health and well-being, that dropped to 17%. For household utility costs, perhaps because Americans are already watching their electricity bills tick upward as data centres proliferate, just 14%. The most pessimistic number of all: only 5% expected AI to improve relations between the United States and China.
“These results tell a clear story,” said Shawn Patterson Jr., a research analyst at APPC. “Americans are paying attention to AI and what they see concerns them. The demand for regulation is not a partisan issue. Majorities across the political spectrum say the government has done too little.”
The Strangest Bipartisan Consensus in American Politics
To appreciate how unusual the AI numbers are, it helps to hold them next to the other issues in the survey. On immigration, the partisan gap between those who thought Kamala Harris would have outperformed Donald Trump and those who didn’t was 75 percentage points. On inflation, 76. On AI regulation, it was 65 points, still large, but the smallest of any policy area tested. More revealing: 24% said Harris and Trump would have done roughly the same job on AI. On immigration, that figure was 7%.
What this suggests is that AI has not yet been claimed by either tribe. It remains contested terrain, uncolonised by the sorting machine that has processed almost every other issue in American public life. Matt Levendusky, a political scientist at Penn, called it “particularly striking,” adding that “concern about AI is bipartisan, and the public is waiting to see what politicians will do. This offers real potential rewards to either party if they can convince the public that they have the correct approach.”
There’s a window, in other words. Whether anyone walks through it is another matter.
Even among people who expected AI to have a very positive impact on the country, 43% still thought the government had done too little to regulate it. The demand for oversight persists even where hope does.
Data Centres and the Anxiety Closer to Home
One of the survey’s more concrete questions asked about something physical, something local: would you support the construction of new AI data centres in your area? The answer, from nearly half of all respondents, was no. Thirty-one percent were strongly opposed. Only 21% were in favour. These are not abstract concerns about superintelligence or algorithmic bias. They are, at least in part, concerns about land use, energy consumption, water usage, noise, and the accelerating sense that something large and consequential is being built nearby without anyone having been properly asked.
Among employed Americans, 41% said they were somewhat or very worried about losing their job, or seeing their hours cut, because of AI. Democrats were more anxious on this score (50%) than Republicans (32%), which is perhaps not entirely surprising given the industries and occupations that tend to cluster around each party. But the anxiety itself crosses the line. It is, like the demand for regulation, broadly shared.
What the survey cannot tell us is what, precisely, people want done. Fifty-two percent say the federal government should take the lead rather than individual states. But the specifics, the licensing regimes, the liability frameworks, the questions about training data and autonomous systems, remain unresolved in public debate. Not because people don’t care, but because the choices are genuinely hard, and the technology moves faster than the institutions meant to govern it.
That gap between public concern and policy clarity is where we all live right now. Whether the bipartisan consensus the Annenberg researchers found translates into actual legislative pressure, or simply dissipates as AI gets sorted into the usual partisan camps, remains one of the more interesting open questions in American public life.
Frequently Asked Questions
Why are Americans more optimistic about AI in medicine than anywhere else?
The survey found 57% of Americans expected AI to benefit medical research, far higher than any other domain. This probably reflects steady coverage of AI-assisted drug discovery, diagnostic imaging, and genomic analysis. Medicine is also an area where people are inclined to welcome any advance that might improve outcomes, even when sceptical of the technology more broadly.
Is it surprising that Republicans also want more AI regulation?
Somewhat, given the party’s general resistance to regulatory expansion. But 53% of Republicans said the government had done too little, compared with 77% of Democrats. Republicans may be worried about different things, such as job displacement or national security concerns about Chinese AI, but the impulse toward oversight converges across party lines.
Why do so many Americans oppose data centres near them?
Data centres are large, energy-hungry facilities that draw heavily from local power grids and, in many cases, significant volumes of water for cooling. Communities have started noticing effects on electricity bills and infrastructure. Opposition likely reflects practical concerns about costs and utilities alongside broader unease about what the AI industry represents for local areas.
Could AI eventually become as politically polarised as immigration or the economy?
Possibly. The Annenberg researchers note that AI has not yet been fully sorted into partisan camps, which is unusual in the current environment. But political issues rarely stay unclaimed for long. As AI becomes more embedded in debates about labour, national security, and regulation, it may well acquire the same partisan valence as other contested issues. Which party moves first may matter a great deal.
ScienceBlog.com has no paywalls, no sponsored content, and no agenda beyond getting the science right. Every story here is written to inform, not to impress an advertiser or push a point of view.
Good science journalism takes time — reading the papers, checking the claims, finding researchers who can put findings in context. We do that work because we think it matters.
If you find this site useful, consider supporting it with a donation. Even a few dollars a month helps keep the coverage independent and free for everyone.