The data: Far more consumers believe AI can already or could one day replace doctors’ medical advice than believe it never could, according to an October eHealth survey of 1,000 US adults.

• 36% think AI can replace doctors' medical advice within the next 10 years.
• 22% believe AI can replace doctors’ medical advice now.
• 30% believe AI will never replace doctors’ advice.

Why it matters: AI is evolving from a source of health information into a tool that shapes consumers’ healthcare decisions. This shift is driven by AI’s constant, low- or no-cost availability when a doctor might not be accessible, plus its ability to quickly answer health questions, incorporate context, and provide detailed follow-up explanations.

That convenience and depth can make consumers more comfortable relying on AI for health advice, potentially viewing it as a substitute for physician guidance. One-third of US adults said they have asked an AI chatbot a medical question in lieu of seeing a doctor, per a September Ipsos poll. Reinforcing that trust, 95% of people who use chatbots for health said the information was at least somewhat helpful, per a June Pew Research Center survey.

The more consumers rely on AI health advice without consulting a doctor—or act on it against a doctor’s guidance—the greater the risk of patient harm. Lawsuits alleging harm from unsafe AI medical advice are increasing.

• In a new case detailed by The New York Times, a family alleges that ChatGPT practiced medicine without a license and advised their daughter, who had severe epilepsy, to taper her medication from 400 mg to 0 without consulting a doctor.
• The chatbot allegedly encouraged her to conceal the plan from her physicians, despite her expressing concerns about tapering at least 158 times, according to the Times' review of her conversations. The woman later died, adding to a growing number of lawsuits against AI platforms alleging wrongful death and other health harms.
• OpenAI said the case involved an early version of ChatGPT that is no longer available.

Implications for AI companies: AI is filling a healthcare access gap, but its guidance shouldn’t be considered as reliable as a doctor’s. Its agreeable, reassuring tone may drive consumers to repeatedly turn to AI with anxiety-driven health questions rather than consult a physician, or even develop treatment plans that contradict medical advice. This risk may be especially pronounced among users who express distrust of doctors or dissatisfaction with their care.

Leading AI companies will likely highlight stronger safety guardrails in newer models. At the same time, OpenAI, Perplexity, Anthropic, and Microsoft have all launched consumer health AI tools, positioning their products not only as sources of health information but also as platforms that generate personalized insights from users’ health data. With some still in preview or limited release, liability concerns could delay broader rollouts or even prompt companies to scale back these offerings.