Quinnipiac Poll: Most Americans Want AI to Slow Down
A new survey from Quinnipiac University shows little public appetite for faster AI. The university's polling operation regularly measures American opinion on political and social issues. This time it found that most US adults want AI development to slow down, and a sizable share want it to stop until safety can be shown.
The results also point to a deeper problem for the industry: most respondents do not trust the people running AI companies.
The headline numbers
The poll asked Americans how fast AI development should move. The answers lean heavily one way:
- 47 percent want the pace reduced.
- 30 percent want development halted completely until its safety has been verified.
- 5 percent want AI to move faster.
Put together, more than three quarters of respondents favor either a slowdown or a pause. Supporters of acceleration are a small minority.
Broad support for independent safety rules
The survey also asked about independent safety standards for AI companies. Proposals of this kind are currently being debated in the US. Support was nearly universal: 86 percent back such standards, even if they slow progress.
That trade-off matters. Respondents were not simply asked whether they like safety. They were asked whether they would accept safety rules at the cost of speed, and a large majority said yes.
Existential worry is mainstream
The poll also measured concern about long-term risks. 73 percent of respondents said they worry that AI could eventually threaten humanity's survival.
Fears about AI ending humanity were once mostly limited to researchers and online forums. This result suggests they are now widely shared among the general public.
A trust gap at the top
The weakest number in the poll belongs to the leaders of AI companies. 74 percent of respondents have little or no trust in them.
Chetan Jaiswal of Quinnipiac University connected the findings directly. "Taken together, these results suggest that the public wants safeguards that do not depend simply on taking company leaders at their word," he said.
Seen together, the results are consistent. People want slower development and independent rules, and they do not trust the executives who might otherwise be expected to regulate themselves.
How the survey was run
Quinnipiac surveyed 1,202 adults in the US between September 24 and 27. The margin of error is plus or minus 3.5 percentage points. Even at the edges of that range, the main findings would not change. Support for slowing down and for independent standards is far larger than the margin of error.
The Bigger Picture
For anyone following AI policy, the main point here is the gap between public opinion and how the industry currently operates. Companies are competing on speed. This poll suggests the public would like them to slow down and to answer to someone other than themselves.
That puts recent developments in a different light. A non-binding code of conduct signed by tech CEOs is exactly the kind of promise that, by Jaiswal's reading, many Americans do not find sufficient. Headlines such as an OpenAI safety lead quitting over company culture are unlikely to close the trust gap either.
It is worth watching whether lawmakers treat numbers like these as political cover for binding rules. Another open question is whether regulatory moves such as the FTC probe into major AI labs gain more momentum as a result. A single poll does not set policy. Still, an 86 percent majority for independent standards is difficult for any policymaker to ignore.