Getty Images/iStockphoto
Can docs integrate AI chatbots into the patient-provider relationship?
Reframing AI chatbots as tools to support shared decision-making can help healthcare professionals engage patients while mitigating the risks associated with them.
2026 was the year the clinical care team expanded, with more and more patients consulting AI chatbots -- not just their providers -- to get healthcare advice.
That's introduced a new dynamic for patients and providers. How can they loop the AI into the patient-provider conversation?
Patients bringing in their own research isn't a new phenomenon. Before AI chatbots, there was Dr. Google, and before that, patients had other resources they could reference.
But patient chatbot use has grown astronomically this year, demanding attention from providers whose patients are influenced by AI information. According to June 2026 figures from KFF, 29% of patients use AI to find health information. That's up from just 17% who said the same two years ago, KFF said.
For some, that spike has raised concerns that AI will peddle medical misinformation or at least stand in the way of the patient-provider relationship.
And those fears might be well-founded.
A September 2026 survey from Bespoke Surgical found that a quarter of surveyed patients said they'd be comfortable with AI replacing their doctor, while 46% said the same about their therapists. A separate August 2026 survey of 1,250 insured adults found that 15% had acted on medical advice from AI without first consulting a doctor.
Healthcare providers acknowledge that AI can have many benefits for patient engagement. However, they insist that there must be a human in the loop. In this case, that means checking in with a provider after getting advice or an explanation from an AI.
But a March 2026 KFF survey found that 42% of patients who use AI to learn about their health don't follow up with a healthcare professional afterward. So actually making that happen -- getting patients to loop their providers in on their AI conversations -- is hard.
Patients might be embarrassed to admit to asking AI about their medical care. Indeed, 2026 data from Zocdoc said that 39% of patients fear their doctor will judge them for using AI, while 38% fear their doctor will not like them using AI and 31% fear their doctor will dismiss them.
AI chatbots will only become more popular. To keep pace with change, healthcare professionals must assess how to talk to their patients about the technology and integrate it as a third voice in the patient-provider interaction.
How can docs get patients to talk about AI use?
Despite patient safety concerns, AI chatbots aren't a provider boogeyman, the data shows. Per the Zocdoc data, 77% of doctors feel positively about their patients' use of AI. It's just that they want their patients to disclose when they use it.
According to Douglas Spotts, an American Academy of Family Physicians board member and Pennsylvania-based doctor, AI can be a valuable facilitator of shared decision-making.
"First and foremost, patients know themselves, but we as physicians know science and medicine, and we're there to really help partner with patients to give them the best advice possible," Spotts said.
AI helps patients parse through their medical needs and preferences before a visit, Spotts added. But it's the job of a healthcare professional to help patients parse through AI information.
"AI can be a helpful tool for preparation, education and follow-up questions, but it should never replace personalized medical advice from your trusted family physician or clinician member of that team," Spotts said. "It's okay to use tools, but I want to be there as the trusted voice of reason to help them sort through those issues."
As a family medicine physician, Spotts is well positioned to have these kinds of conversations. Because his relationships with patients are long-term, he has naturally cultivated trust and rapport that helps him broach the topic of AI use.
Spotts starts by simply asking patients what they'd like to talk about during their visit that day, or another similarly open-ended question.
"And then I go a little bit further," he continued. "I'll say, 'What have you looked at online? What have you heard from influencers? What have you heard from AI? Have you used any of those resources?' And I just try to make it very comfortable for them to know that it's okay to ask questions and okay to gather information."
This tactic is useful when patients bring in a polished list of questions about recent labs, but it works double-duty when patients bring in medical misinformation they picked up from AI informed by faulty sources.
Take, for example, the conflicting messaging around vaccines. Spotts said he -- and the AAFP, for that matter -- operates with the understanding that vaccine science hasn't changed. When Spotts assures patients that it's okay to seek out information on their own and take control of their own health, he can open the door to a more collaborative discussion around hot-button topics like vaccines.
"We can definitely lead them through the guidance of where we get our information and what it's based on: evidence and science," Spotts said.
The data backs up Spotts' strategy. According to the 2026 Edelman Trust Barometer, patients don't want to hear why they're wrong.
Even if their provider is offering medical advice they doubt, patients want to know that their values, lifestyle and resources are being respected. Patients said they don't want their providers to prove them wrong, but rather to demonstrate their knowledge of the patient's community and work collaboratively with them to reach a consensus.
Time constrains conversations about AI chatbots
It takes time to discuss AI chatbots with patients -- but that is something providers famously don't have. Even if it's just level-setting that it's okay to use an AI chatbot, that conversation might take away from another topic during the visit.
According to the Zocdoc survey, 83% of providers said they need to correct AI-generated information patients take in, while 63% said this adds time to appointments.
But there is an upside to bringing AI into the patient visit. Ambient scribing, for example, has allowed Spotts to shift his focus away from documentation and reprioritize his communication with patients.
"All family doctors know you can't cover everything in every visit, so you want to build up that time to do what is important to the patient and take care of the need then," he added. "And if there are other issues, you bring them back."
That follow-up can even happen asynchronously, such as through the patient portal, Spotts recommended. And with AI being embedded in many patient portal tools, clinicians can focus on more complex messages and leave the technology to handle things like scheduling or prescription refills.
Still, there's a reason patients use AI in the first place, and it's usually because cost and appointment availability limit access. In these cases, AI isn't additive to their healthcare encounters; it is a way to circumvent their biggest barriers.
Until healthcare can address these fundamental patient engagement pitfalls, patients and providers will be unable to realize the benefit AI brings to their interactions.
Sara Heath is an executive editor at Xtelligent Healthcare Media, where she covers patient engagement, healthcare policy and health IT.