Picture this: you are sitting at your kitchen table, and a strange health question has been on your mind. You could call your doctor, but it’s late. You could search the internet, but you know the results will be a jumble of ads, old forums, and contradictory articles. So you open a chatbot instead. You type a few words, and within seconds it gives you a calm, clear answer. It sounds confident. It sounds helpful. But then you pause: How do I know this is true? That moment of hesitation is not just a personal quirk—it reflects where the United States is as a country right now. According to the latest KFF Health Misinformation Tracking Poll, Americans have welcomed artificial intelligence into their daily lives with remarkable speed, but they have not yet decided whether they can trust it. Most adults say they have used or interacted with AI, and many are beginning to rely on it for ordinary tasks, from writing emails to planning meals. Yet when the conversation shifts to something as important as health, confidence collapses. A majority of adults say they are not confident they can tell the difference between what is true and what is false when they get information from an AI chatbot. Even among people who already use these tools, about half say they cannot reliably tell fact from fiction. This is not a story about technology being loved or hated. It is a story about people trying to make sense of a powerful new tool that has arrived faster than society’s ability to understand it. The stakes are especially high because AI is already embedded in the search engines and social media platforms we use without thinking. Many people are likely interacting with AI without even knowing it, which makes the question of trust all the more urgent. The poll makes one thing clear: Americans are not ready to hand over their most important questions to machines, but they are also not turning away. They are simply unsure, cautious, and watching closely.
That uncertainty is not for lack of exposure. The KFF poll found that about two-thirds of adults have used or interacted with AI, and about one-third say they do so at least a few times a week. For many, it has become routine. Eleven percent of adults say they use or interact with AI several times a day, which means artificial intelligence is now woven into the rhythm of ordinary life. At the same time, about a third of adults say they never use or interact with AI, though that number may be misleading. As AI becomes built into internet search engines like Google and social media platforms like Facebook and Instagram, people can benefit from or be influenced by AI without ever recognizing it. If you have ever asked a search engine to summarize an answer, let a social media app suggest a caption, or followed an automated customer service conversation, you have likely brushed against AI without calling it by name. The poll’s question captures only deliberate interactions, so the true reach of AI is probably even broader. Not surprisingly, younger adults are more likely to say they use AI regularly. Nearly half of adults ages 18 to 29 say they use or interact with AI at least several times a week, compared with much smaller shares among older adults. Education also matters: adults with a college degree are more likely to use AI weekly than those without one. These patterns make sense, because younger and more educated people tend to encounter new technologies in school, at work, and through their social circles. But the poll shows that AI is not just a toy for tech enthusiasts anymore. It has become a fixture in American life, and it is reaching people of all ages, backgrounds, and levels of digital skill. The fact that so many people are using AI while simultaneously doubting it says something important about this moment. We have not yet learned where the boundaries are. We use the tool, we appreciate its convenience, and then we worry about what it might be getting wrong.
That worry is especially clear when people are asked whether they can tell truth from falsehood in what AI chatbots tell them. The KFF poll found that only about four in ten adults are confident they can tell the real from the invented, the accurate from the misleading, when they read information generated by a chatbot. Specifically, nine percent of adults say they are very confident and another 33 percent say they are somewhat confident. That leaves a majority—56 percent—saying they are not too confident or not at all confident in their own ability to judge what is true. Perhaps the most striking finding is that confidence does not rise much among people who actually use AI. Even among those who have used or interacted with AI, half say they are not confident that they can tell fact from fiction in chatbot responses. That suggests that familiarity with the technology does not automatically make people feel more equipped to evaluate it. To be fair, AI users are more likely than non-users to feel some confidence: 49 percent of users say they are at least somewhat confident, compared with 32 percent of people who have never used AI. But that still means a huge number of active users are flying without a clear sense of what to trust. Age plays a role in how people see their own abilities. Younger adults are more likely than older adults to feel they can spot false information from AI chatbots. Seven in ten adults ages 65 and older say they are not confident they can tell what is true from what is false on these platforms. Even among AI users, younger adults are more confident than those over 50. But the overall picture is sobering. People are being asked to make decisions about what to believe, but they do not feel equipped to do so. Imagine being given a new tool that can tell you almost anything, but no instruction manual that tells you when it might be lying. That is where many Americans say they find themselves. They are not rejecting the tool, but they are deeply uncertain about how to use it without being misled.
Nowhere is that uncertainty more consequential than in health. The KFF poll asked Americans whether they use AI chatbots to find health information and advice, and the answer is that many of them do. About one in six adults say they use AI chatbots at least once a month for health-related questions. Among younger adults ages 18 to 29, that number rises to one in four. That is a meaningful shift in how people are seeking medical guidance. Instead of reading a WebMD article or waiting for a doctor’s appointment, a growing number of people are asking machines to help them understand symptoms, medications, treatments, and lifestyle advice. But the public is nowhere near convinced that these tools are accurate. Only five percent of adults say they are very confident that health information from AI chatbots is accurate, and 31 percent say they are somewhat confident. The other six in ten are skeptical, saying they are not too confident or not at all confident. Even among people who use AI, 56 percent say they are not confident that the health information they get is accurate. That is a remarkable sign of maturity from the public: they are using the tools, but they are not swallowing everything they read. The poll also found that Black and Hispanic adults, as well as adults under 50, are somewhat more likely than older adults and White adults to say they have confidence in AI health information. But even among those groups, a majority remain unconvinced. There is something important about this skepticism. Health information is not like a recipe or a history fact. If a chatbot tells you the wrong temperature for baking chicken, you might ruin dinner. If it tells you to skip a certain medication or ignore a possible warning sign, the consequences can be far more serious. People seem to know this instinctively. They may trust AI to draft a note or summarize a document, but when it comes to their bodies, they are hesitant. That hesitation is rational. The challenge for health professionals and public health organizations is to help people find reliable sources and to teach them how to think critically about AI-generated health advice, rather than assuming that people will simply accept whatever a chatbot tells them.
The public’s cautious attitude becomes even clearer when people are asked what they trust AI to do. Roughly half of adults say they trust AI chatbots to provide reliable information about practical tasks like cooking and home maintenance, and about the same share trust them to provide reliable information about technology. But trust drops sharply when the topic becomes more serious. Only 29 percent of adults say they trust AI chatbots to provide reliable health information, and only 19 percent trust them to provide reliable information about politics. This is a consistent pattern: people seem to see AI as a useful assistant for ordinary, low-stakes problems, but not as a credible source for questions that involve their health or their role as citizens. Trust is higher among people who use AI, but even among users, only 36 percent trust chatbots for health information and only 24 percent trust them for politics. In other words, even the people most comfortable with this technology draw a line at the same place. They are willing to ask a chatbot how to remove a stain from a carpet, but they are much less willing to ask it whether a symptom is dangerous or whether a political claim is true. This pattern makes sense from a human perspective. When a mistake is easy to correct or does not matter much, we are more willing to take a chance on a machine. When the cost of being wrong is high, we become cautious. The poll suggests that Americans have not lost their common sense in the rush to embrace new technology. They understand, perhaps without even putting it into words, that AI can be confidently wrong. It can present misinformation in clear, persuasive language. It can mix fact and fiction in ways that are hard to catch. So they use AI where it feels safe, and they hesitate where it does not. That is not ignorance. That is discernment. The problem is that many people do not feel confident in that discernment, and they are right to worry, because distinguishing a reliable AI answer from a fabricated one requires skills that most of us have not yet developed.
For all the attention on AI’s potential to help or harm, the most common reaction is not enthusiasm or fear—it is uncertainty. When asked whether AI is doing more to help or more to hurt people who are trying to find accurate health information online, only about one in five adults say it is helping and a similar share say it is hurting. The majority, 55 percent, say they are not sure. Even among people who use AI, about half are not sure whether the technology is making it easier or harder for people to find trustworthy health information. Young adults are divided, with about three in ten saying AI helps, three in ten saying it hurts, and the rest unsure. Older adults are even more hesitant. This is what an emerging technology looks like in the eyes of the public: a lot of promise, a lot of danger, but not enough evidence to settle the question. The human mind craves clarity, and this poll delivers none, because the reality does not yet have clarity. AI is already changing the information environment in ways both good and bad. It can provide quick, accessible health information to people who might otherwise go without. It can also spread misinformation at scale, with a tone of authority that makes falsehoods look like facts. No one knows yet which force will win out. That is why so many Americans are withholding judgment. It is not indecision born of ignorance. It is the sensible response of people who have watched technologies come and go, and who know that tools can be used well or used badly. The KFF poll is ultimately a reminder that technology does not have to be either a savior or a villain. It can be something we learn to live with carefully. Most Americans seem to understand that. They are using AI, testing it, doubting it, and watching what happens. They are not ready to trust it with their health, and they are not sure it will make the world better, but they are not turning away either. That leaves the rest of us with a challenge: to help people build the skills they need to navigate this new landscape, to hold technology companies accountable for accuracy, and to make sure that in the age of machines, something as human as good health is not left to chance.