TL;DR — Key Takeaways
- A Quinnipiac University poll found 77% of U.S. adults favor slowing or stopping development of powerful AI systems until their safety can be evaluated.
- Eighty-six percent support independent AI safety standards even if they slow development, while 69% also consider keeping pace with China important.
- Opposition extends to AI infrastructure, with 72% saying they would oppose an AI data center in their community, up from 65% in Quinnipiac’s March poll.
Anthropic CEO Dario Amodei recently made headlines by calling for frontier AI development to slow down over safety concerns. OpenAI CEO Sam Altman and several other AI leaders backed the idea, while others have pushed back on an industrywide slowdown. A new Quinnipiac University poll suggests most Americans favor the more cautious approach: 77% said companies should slow or stop development of powerful AI systems until their safety can be evaluated.
That 77% included 47% of adults who supported slowing development of frontier systems while their safety is assessed, and 30% who said development should stop altogether. Another 14% favored continuing at the current pace and 5% wanted companies to move faster. Separately, 86% supported requiring AI developers to meet independent safety standards even if doing so slows development, and 91% said that establishing guardrails for AI systems is important.
Those views coexist with concern about international competition. The poll found that 69% of respondents said it is very or somewhat important for the United States to keep up with China in AI development. Quinnipiac said 65% of respondents hold both views, simultaneously considering keeping pace with China and establishing AI guardrails as important.
“The public is not neatly divided into a competition camp and a safety camp,” said Chetan Jaiswal, professor of computer science and associate chair of the Department of Computing at Quinnipiac University. “That does not tell us how people would resolve every trade-off, but it does show that valuing competition should not be mistaken for opposing safeguards.”
The poll’s results also show limited trust in the people leading AI companies. Seventy-four percent said they had either “not so much” trust or no trust at all in AI company leaders, compared with 24% who reported some or a lot of trust.
Meanwhile, 53% said AI will do more harm than good in their day-to-day lives. That was roughly unchanged from 55% in Quinnipiac’s March survey. And 73% said they were very or somewhat concerned that future AI systems could threaten human survival. But when asked to choose between two sources of potential harm, 52% were more concerned about humans using AI to harm humanity, compared with 33% who were more concerned about autonomous AI agents.
The industry also faces growing resistance to the infrastructure required to support AI workloads. Seventy-two percent said they would oppose an AI data center in their community, up from 65% in the March poll. Support for data centers fell from 24% to 21% over the same period.
Those findings track closely with other recent polling. Similar to Quinnipiac, a Gallup survey conducted in March found that 71% of Americans opposed construction of an AI data center in their area. In a follow-up survey in April, opponents most often cited resource use, including water and energy consumption. Additionally, a recent Research Center survey found that attitudes toward data centers had become more negative since January, with half of Americans saying they are mostly bad for home energy costs and 49% saying they are mostly bad for the quality of life of nearby residents.
Quinnipiac surveyed 1,202 U.S. adults from Sept. 24 through Sept. 27 using live telephone interviews. The poll has a margin of sampling error of plus or minus 3.5 percentage points, including the design effect.
