# How to ask AI about your health: the 9 rules I use An open-ended health question is just Dr. Google with better grammar. Give your AI rules instead, and you get signal instead of noise. Here are the 9 I use every time. ![Article illustration](https://kevinwimer.com/media/3eb33be6-23d9-4d21-b3a0-8f28b37cf8a1) Author: Kevin Wimer Canonical URL: https://kevinwimer.com/cancer/resources/ask-ai-about-your-health/ Stable ID: 5209b08a-d248-40cd-bad2-dc8bf8518a23 Revision: dcbeb913-d673-4bc6-aa75-6ed8df652bc0 Published: 2026-10-27T18:11:21.665Z Updated: 2026-10-27T18:11:21.665Z Reviewed: 2026-10-27 ## Who this is for For patients and caregivers who use an AI chat tool to understand labs, scans, symptoms or treatment options, and want clearer, more honest answers to bring to their care team. It's easy to use AI for your health like a Magic 8 Ball. Vague question, vague answer, feel better or worse, close the app. I was diagnosed with lung cancer in July 2024. I never smoked. I've spent thousands of hours using AI since, and the biggest thing I've learned is simple: an open-ended health question is just Dr. Google with better grammar. One lab is off, AI lists five causes, and you're up at 2 AM over a nothingburger. I don't need help finding new things to worry about. Cancer covers that. So I don't ask open questions anymore. I give it rules. ## The 9 rules 1. **Be candid.** No sugarcoating. 2. **Make a call.** What would you pick for a loved one? 3. **Quantify it.** Marginal or substantial? 4. **Give odds.** Realistic ranges, no fake precision. 5. **Challenge me.** Look for why I'm wrong. 6. **Fact vs speculation.** What we know, what we don't. 7. **Skip the boilerplate.** Flag only what's urgent. 8. **Next steps, in order.** Who to call, what to ask. 9. **Define success.** How will I know it's working? **Open-ended questions get you noise. Rules get you signal.** ## Why these work for me Every rule is there to cut something I don't want. "Be candid" and "Make a call" cut the hedging. I'd rather hear the hard version now than piece it together later. "Quantify it" and "Give odds" stop the vague maybes, without letting it pretend to know more than it does. "Challenge me" is there because if I'm wrong, I want to find out from my AI, not at my next scan. "Fact vs speculation" and "Skip the boilerplate" are what keep me from that 2 AM spiral. And the last two turn an answer into a plan: who to call, what to ask, and how I'll know if it's working. ## What the rules don't do They don't make AI my doctor. Every answer gets checked. The good ones go to my oncologist as questions, not as orders. **AI doesn't make my decisions. It makes sure I'm making them with everything on the table.** ## Make the rules go further Rules get you a better answer. Context gets you an answer about you. - [Connect your real records with ChatGPT Health](/cancer/resources/chatgpt-health-medical-records/). An AI that can't see your chart is guessing. - [Give your AI a notebook](/cancer/resources/ai-health-notebook/) so it knows what your doctors said, not just what's public. - [Ask about your data, not your symptoms](/cancer/resources/ask-ai-about-your-health-data/). The same rules work on years of wearable data. - [Use AI to do the paperwork and research](/cancer/resources/ai-cancer-research/) - [AI and your care](/ai-your-care/) I'm not a doctor and this isn't medical advice. It's what's working for me. Please talk to your care team before you change anything. ## Limits and context One patient's personal method, not medical advice. Rules make answers clearer, not correct. AI can still be wrong, so check what matters with your care team. ## Relationships and disclosures No paid placement. I have no investment, advisory role, paid partnership or other financial relationship with any AI company or product. ## Sources - [AI ideas, specialist validation](https://x.com/kevinace/status/2111977278465024129) — Kevin Wimer on X; experience; checked 2026-10-27; Personal experience, read in Kevin's supplied X export. Not independent clinical evidence. Topics: ai-your-care, getting-started Capabilities: Related record IDs: af2e7c15-d7c9-4b91-af55-0a9a764eb515, 9f20f130-cad1-494e-bf40-bd02d30f01cd, 2a74a16b-3c6d-458f-a871-c95e0b847f58, eb3bb171-b7b6-4a5e-9466-bcab2b26f1c9, a2420711-3cbc-4edc-868b-da3c4dcd5bef Follow Kevin on X: https://x.com/kevinace Educational information, not an individualized treatment recommendation.