1 in 3 adults now turn to AI for health advice

Approximately one third of U.S. adults have turned to artificial intelligence (AI) chatbots for health information in the past year, putting AI on par with social media as a go-to source for medical questions.

The finding comes from a new KFF Tracking Poll on Health Information and Trust, released March 25. Of 1,343 adults surveyed between Feb. 24 and March 2, 29% had sought information about their physical health from an AI tool in the past year and 16% had done so for mental health concerns.

Overall, 32% had used AI for some form of health advice — on par with the 29% who said they turned to social media and trailing only internet search engines (68%) and health care professionals (80%). It should be noted that many search engines now surface AI-generated summaries, meaning exposure to AI health content is likely higher than self-reported use suggests.

In recent months, several major technology companies have rolled out health-specific AI products, including ChatGPT Health and Amazon Health AI, pitched as personalized tools that can interpret lab results, explain diagnoses and help patients prepare for visits.

Related content: ChatGPT Health missed half of medical emergencies in first independent safety test

Younger, uninsured and lower-income patients lead the shift

Use of AI for health advice skews heavily toward younger adults. More than one-third (36%) of those aged 18 to 29 used AI for physical health questions in the past year, and 28% used it for mental health, which is roughly three times the rate of adults 50 and older.

Much of that behavior has been attributed to access and affordability. Among users under 30, 38% said not having a regular physician, or being unable to get an appointment, was a major reason they turned to AI, and 29% cited cost. Among users with household incomes below $40,000 annually, 32% cited affordability as a major reason.

Uninsured adults were more than twice as likely as insured adults to use AI for mental health advice (30% vs. 14%), and Black (21%) and Hispanic (19%) adults were more likely than White adults (12%) to do the same.

Sixty-five percent of users said a desire for quick or immediate information drove them to AI. Another 41% wanted to look up information before deciding whether to see a provider, and 36% said they felt more comfortable looking things up privately.

Many never follow up with a clinician

A substantial share of patients stop at the chatbot.

Among adults who used AI for physical health advice, 42% did not follow up with a human clinician. For mental health, 58% skipped the follow-up, and younger adults were roughly twice as likely as older adults to skip the follow-up.

Amber Maraccini, Ph.D., CPXP, vice president of health care and life sciences at Medallia, said in a recent conversation with Medical Economics that the shift is less about replacing clinicians than about how patients now arrive at appointments.

"For the first time, patients are able to encounter meaning before they actually encounter their clinician," Maraccini said. "This is a really big deal … but it's also risky, because there's an emotional reality that we have to keep in mind with the information."

Maraccini said the instinct for some physicians may be to push back when a patient walks in holding an AI-generated summary, but she argued that is the wrong move. Patients have been bringing outside information into visits for years in the form of printouts, Google searches and app screenshots. AI is simply the newest version.

"The most productive path would not be to dismiss the effort," she said. "Instead, it's looking at the information and saying, ‘Let's look at this together.’" The clinician, she noted, has context the chatbot does not: the patient's history, comorbidities and goals. “When you, as a clinician, are showing that you're not just an interpreter of data, it actually adds to your credibility and trustworthiness.”

Are there privacy concerns? Sure

The poll noted a notable disconnect on privacy. Roughly 77% of adults said they are concerned about the privacy of personal medical information provided to AI tools. Yet, among adults who used AI for health in the past year, 41% said they had uploaded personal medical information — test results, doctors' notes or similar records — into a chatbot to get personalized explanations. That works out to 13% of all U.S. adults, rising to 19% of those ages 18 to 29.

Even among adults who had uploaded personal medical data, 65% still said they were concerned about the privacy of that information.

Satisfaction with AI responses was high across the board. KFF found 92% were at least somewhat satisfied with AI responses on physical health questions and 85% on mental health, though only 19% and 27%, respectively, said they were “very satisfied.”

Trust in AI among non-users remained low, as just 18% of adults who had not used AI for physical health information said they trust the tools to provide reliable health information.

The problem is deeper than misinformation

At an American Public Health Association event held April 6 during National Public Health Week, Jessica B. Steier, Dr.P.H., CEO of Unbiased Science, delivered a keynote address arguing that the larger misinformation challenge in American health care is fundamentally about trust rather than data.

"The misinformation crisis is not primarily a knowledge problem. It's a trust problem," Steier said during the event. She urged clinicians and public health advocates to lead with listening rather than correction.

Related content: Health care epidemic of misinformation is rooted in problem of trust, experts say

Aerial Petty, D.O., a family physician and vice president of the Young Leaders Council of National Medical Fellowships, said during a panel discussion at the same event that patients spend the overwhelming majority of their lives outside the clinic, gathering information on their own. "It's important to meet people where they're at and understand where they're coming from," she said.

For Maraccini, the warning sign physicians should watch for isn’t patients using AI — it’s patients saying they prefer it to their physician.

"A red flag is when the patient [is] saying it's easier to ask AI than my doctor … or AI is more empathetic than my doctor," she said. "To me, that's more of a red flag on the relationship and the communication skills with the provider."

Rather than steering patients away from the tools, she suggests that clinicians ask how they are using them and help them use them more effectively, particularly for visit prep and post-visit reinforcement of care plans.

"Instead of clinicians trying to shy away or avoid those conversations," Maraccini said, "they should be leaning in proactively and teaching their patients how to best use the technology."