One-third of adults are using AI chatbots to seek health information

AI chatbots are hitting the mainstream with consumers who are looking for faster and more convenient methods of accessing health information. 

The latest Tracking Poll on Health Information and Trust from KFF finds that one in three US adults (32%) is now using a chatbot like ChatGPT, Claude, or Google Gemini to get physical or mental health advice. 

About one in five of those users said they’re turning to AI because they don’t have an established relationship with a healthcare provider, and an additional 19% use AI because they simply cannot afford to access traditional healthcare. Younger respondents and those in low income households were among the most likely to cite these reasons. 

Other key insights from the survey include: 

• AI is becoming a go-to source for mental health support, with more than a quarter (28%) of people aged 18-29 saying they’ve used AI for information about their mental health or emotional wellbeing. Uninsured adults, as well as those who identify as Black or Hispanic, are also significantly more likely than other groups to turn to AI for mental health reasons. 

• Many users feel that AI helps with building knowledge and confidence when addressing health issues, with 41% wanting to look up information before seeing a provider and 36% feeling more comfortable looking up health questions privately. 

• Using AI helps to build trust in the tools, with much higher numbers of active AI users (69%) stating they trust the answers they’re getting about physical health compared to about 18% of people who have never used AI to research physical health concerns. Mental health followed a similar pattern (62% of users trust AI for mental health vs. only 16% of non-users). 

• Most (77%) respondents say they’re concerned about the privacy of personal health information provided to AI companies, but that’s not stopping some from uploading their data anyway. Just over 40% of those who have used AI for physical or mental health (equating to 13% of all adults) have uploaded specific medical information to an AI tool. 

• AI users often don’t follow up with a healthcare provider after engaging with a chatbot. Only 58% of users say they connected with a professional after looking into a physical health issue. Even fewer reached out to a healthcare provider with a mental health concern, with just 42% of users following up with a real person. Younger adults are more than twice as likely as older people to say they did not follow up with a human after a mental health query to a chatbot.

The data clearly illustrates that consumers believe AI tools can fill a need that hasn’t been adequately addressed by the traditional healthcare system: simple, fast, and seemingly reliable access to answers about health concerns. 

And unlike WebMD and regular internet searches, chatbots can personalize the answers, help guide users down specific pathways of interest, and even actively provide data-backed reassurance that the worst-case scenario isn’t necessarily the most likely one – all without the user having to jump through hoops to have an asynchronous conversation on a portal or “bothering” a provider with a barrage of questions. 

The potential benefits are many – or they would be, if the technology was mature enough to deliver on its promises in a reliable, accurate, and intelligent manner. But it’s just not there yet. 

Despite an explosion of consumer-focused healthcare tools and platforms over the past few months, generative AI is still not fully reliable for even the most basic healthcare queries. Several recent studies have highlighted significant issues with generative AI tools, including consistent under-triaging of potential emergency situations and susceptibility to absorbing and repeating false health information. 

The use of chatbots for mental health has also raised red flags, particularly after a string of lawsuits alleging that AI tools played an active role in encouraging or enabling suicides.   

For example, one report from researchers at Stanford University and Common Sense Media states that there is an “unacceptable risk” for users under the age of 18 to engage with AI chatbots as social companions, as younger teens may be likely to discuss their emotional or mental health with these tools instead of real people and receive replies that could validate impulses toward self-harm. 

With the KFF survey confirming that younger people are more likely to treat chatbots as a personal confidante while being less likely to follow up with a care provider about mental health concerns, there is a growing signal that the tech industry needs to work with policymakers and clinicians to set stronger boundaries around the use of AI tools for these purposes. 

Some efforts are already under way, with industry groups, state legislatures, and AI companies themselves trying to define and address the risks. But their work is already lagging behind the adoption curve. With chatbots quickly going mainstream for health, it will be imperative for these processes to get faster and more comprehensive in order to maximize the benefits of AI for healthcare while avoiding the pitfalls. 

Jennifer Bresnick is a journalist and freelance content creator with a decade of experience in the health IT industry.  Her work has focused on leveraging innovative technology tools to create value, improve health equity, and achieve the promises of the learning health system.  She can be reached at [email protected].