AI can translate an MRI report into plain English, summarize your test results, and answer a medication question with total confidence. The problem is that confidence is false, and in healthcare, the cost of a bad answer can be huge. I’m Heather Johnston, MD, and I’m living the patient side right now: heart failure that demands consistent cardio, plus a hip injury and other orthopedic issues that make the gym feel like a minefield. After getting vague guidance from real clinicians who were focused on their own “slice,” I tried something different. I uploaded my MRI reports and heart testing into Claude, got a customized routine, and then did the most important step: I asked my physical therapist to sanity-check it before I followed it. From there, I zoom out to how other patients are using AI chatbots like ChatGPT, Claude, and Gemini for healthcare advice, symptom searches, test interpretation, and insurance questions. I break down what large language models actually are, why the specific model you’re talking to matters, and what the latest research says about medical accuracy, user error, and the risk of misleading answers delivered in a polished tone. I also dig into privacy: most chatbots are not covered by HIPAA, which changes the stakes when you share sensitive health information. Then I’m joined by viral TikTok creator Ty Eveland, whose “AI doctor” trolling with baby Tuna is hilarious and deeply instructive. His clips show how bots can miss sarcasm, context, and reality, while still sounding authoritative and confident. I finish up with practical, safer ways to use AI as a patient today, plus what tools like ChatGPT Health may mean for the near future. Subscribe, share this with a friend who asks chatbots for medical advice, and leave a review with your biggest AI-in-healthcare question.
