TECH | POLICY
Nearly two-thirds of U.S. adults say artificial intelligence is developing too quickly. The deeper concern is not just speed — it is whether people, workers and institutions will retain authority as AI becomes more powerful.
Nearly two-thirds of Americans now say artificial intelligence is developing too quickly, offering another sign that public unease is growing even as companies and governments race to expand the technology. Sixty-four percent of U.S. adultssay AI is developing too fast, compared with 27% who describe the pace as about right and just 8% who say development is too slow, according to an AP-NORC poll released October 8. The concern crosses party lines, although not evenly: 76% of Democrats and 54% of Republicans say AI is moving too quickly. The nationwide poll surveyed 2,140 adults between September 24 and 28 and has a margin of sampling error of plus or minus 2.9 percentage points.
The Concern Goes Beyond Speed
The more consequential finding may be what respondents want government to protect. About eight in 10 adults say keeping AI under human control should be an extremely or very important government priority, and a similarly large share wants policymakers to prioritize protecting U.S. workers. Those views are broadly shared by Democrats and Republicans, even while the parties diverge more sharply over environmental concerns and international competition. That suggests the public debate is moving beyond whether AI is useful. Americans are increasingly asking who retains decision-making authority when AI enters workplaces, schools, health systems, financial services and other consequential parts of daily life.
This is not an isolated poll result. A Pew Research Center survey released in June found nearly the same thing: 63% of adults said AI was advancing too quickly, while only 19% believed it was moving at the right pace and 2% thought it was moving too slowly. Pew also found 67% of Americans had little or no confidence in the federal government’s ability to regulate AI effectively, while roughly six in 10 lacked confidence that U.S. technology companies would develop and use the systems responsibly. The consistency between the surveys matters. Public concern does not appear to be a reaction to one controversial product launch or headline. It increasingly looks structural.
People Want AI — But Not Without Guardrails
That distinction is important because skepticism about speed is not the same as opposition to technology. The AP-NORC poll found that many Americans still consider technological innovation and economic growth important government priorities. What they appear less willing to accept is the idea that deployment itself should dictate the rules. Earlier AP-NORC research found only 9% of adults were extremely or very confident that information from AI systems, chatbots or AI-powered search was reliable and factual, while 49% had little or no confidence. Just 20% said verifying AI-generated information was easy. The tension is increasingly clear: people are already encountering AI in ordinary systems while remaining unsure whether they can verify what those systems produce or challenge what they decide.
That gap becomes more consequential as AI moves from answering questions to influencing outcomes. Human control can mean very different things depending on where the technology is deployed. In hiring, it could mean requiring a person to review a candidate rejected by an automated screening tool. In lending or insurance, it could mean preserving a meaningful appeals process when an algorithm contributes to a denial. In education, health care or government benefits, it could require disclosure when automated systems are making recommendations and a clear path to human review. The principle sounds straightforward; the policy choices needed to enforce it are not.
Washington Has a Different Incentive
The public’s emphasis on restraint also arrives while federal policy increasingly emphasizes technological leadership. The administration has framed advanced AI development as an arena of economic and geopolitical competition, including competition with China, while encouraging accelerated American development. Yet the AP-NORC poll found slightly fewer than half of adults considered competing with other countries an extremely or very important AI priority, compared with the roughly eight in 10 emphasizing human control and worker protection. That creates a widening policy tension: government and industry frequently discuss how quickly the United States can develop AI, while the public appears more concerned about the conditions under which those systems are allowed to operate.
Confidence that political leaders can resolve that tension is limited. Only 31% of adults approve of Donald Trump’s handling of AI, while 23% say Democrats would do a better job managing the technology and 21% favor Republicans. The largest group — 39% — trusts neither party to handle AI well. That may be the most important number in the survey. Americans are not merely expressing anxiety about a technology moving quickly; many also doubt the institutions responsible for setting its boundaries.
The next phase of the AI debate therefore may not hinge on whether another model becomes faster, cheaper or more capable. The harder question is whether governance can become specific enough to meet what the public is asking for: human review, worker protections, accountability when systems fail, transparency about automated decisions and meaningful ways to challenge consequential outcomes. Americans appear increasingly willing to use AI while simultaneously questioning whether anyone has established where its authority should stop. The technology may continue moving quickly. The unresolved issue is whether the rules around it can move quickly enough to keep people in control.