Americans worry ai is moving faster than safeguards in health care

A new AP-NORC poll suggests Americans are uneasy about the pace of artificial intelligence, and that concern extends into the health care conversations now surrounding the technology. Most adults surveyed said AI is developing too fast, while far fewer said its pace is about right or too slow.

That reaction does not mean people reject AI outright. It does suggest that many want stronger guardrails before the technology becomes more deeply embedded in daily life, including in medical settings where errors, bias, privacy problems, or unclear oversight could affect patient care.

What the poll found

The AP-NORC survey found that 64% of U.S. adults think AI is developing too fast, 27% say the pace is about right, and 8% say it is moving too slowly. The poll also found broad concern that the government should make sure AI stays under human control and that workers are protected as the technology spreads.

Americans were also skeptical about political leadership on AI. Few respondents said either major party has a clear advantage on the issue, and the survey found low approval of the president’s handling of AI. In other words, this is not just a technology story; it is also a trust story.

Why this matters for health care

AI is already being used in health-related settings, from imaging support to workflow tools and risk flagging systems. The Food and Drug Administration says it has authorized more than 1,600 AI-enabled medical devices for marketing in the United States as of September 2026. The agency also says it is actively seeking feedback on how to regulate generative AI-enabled medical devices.

That matters because public confidence can shape whether patients, clinicians, hospitals, and regulators are willing to use these systems. If people believe AI is moving ahead faster than oversight, they may be less likely to trust it when it appears in a clinic, emergency department, or screening program.

What the evidence actually shows

The poll measures opinion, not safety or clinical performance. It does not show that AI is harmful in all health applications, nor does it prove that a slower pace would automatically improve outcomes. It does, however, show that many Americans want caution.

Federal health agencies are taking a similar approach. The FDA says AI and machine learning have the potential to transform health care, but it also emphasizes careful management across the full product life cycle. Its recent discussion paper on generative AI-enabled medical devices focuses on risk assessment, premarket evaluation, and postmarket monitoring. That is a reminder that even when AI has promise, regulators still need to test, monitor, and update guardrails.

NIH sources also describe both promise and limits. NIH has reported that AI may help doctors diagnose patients faster, but researchers have also found that models can make mistakes in how they explain medical images or reasoning, even when they reach a correct answer. For health care, that distinction matters: a system can appear helpful and still produce unreliable explanations or unsafe recommendations.

What remains uncertain

One open question is how much trust people will place in AI once they see it used in specific medical tasks. Public opinion about the technology may change if AI is associated with better access, shorter wait times, or more accurate analysis. It could also worsen if people hear about errors, data privacy breaches, or systems that are difficult to audit.

Another unknown is whether current oversight keeps pace with rapid product development. The FDA says it is working on regulation and guidance, but generative AI is still a fast-moving area. Even well-meaning systems can drift in performance, behave differently after updates, or reflect weaknesses in the data used to train them.

There is also a difference between using AI for low-risk support tasks and using it for higher-stakes clinical decisions. A scheduling assistant or documentation tool is not the same as a system that helps guide diagnosis, triage, or treatment. Public concern may rise as the stakes rise.

Practical context for patients and families

If AI shows up in a health setting, it is reasonable to ask a few straightforward questions: What is the tool used for? Who reviews its output? Has it been cleared or authorized by the FDA if it is a medical device? How is patient data handled? What happens if the tool is wrong?

Patients do not need to become technical experts to make use of these systems. But informed questions can help make sure AI is assisting care rather than quietly replacing human judgment where it still matters most.

For clinicians and health systems, the message from the poll is also useful. Patients may accept AI more readily when they understand its role, know that humans remain accountable, and can see that the technology is being used carefully rather than aggressively.

The bottom line

The AP-NORC poll suggests Americans are uneasy about the speed of AI development, and that unease is highly relevant to health care. AI already has a place in medicine, but the question for many people is not whether it can be used. It is whether it can be used safely, transparently, and with enough oversight to earn trust.