Judgment in the Age of AI · Article 9
AI can help organise a health question, but it cannot take responsibility for diagnosis. A response may lack medical history, drug interactions, test results and physical observation; the same symptom can also have very different levels of seriousness. The safest use prepares you for care rather than deciding whether care is needed.
Divide tasks into green, amber and red
| Level | Appropriate AI use |
|---|---|
| Green: general knowledge | Explain common terms, organise information already given by a clinician, and prepare non-urgent questions. |
| Amber: individual advice | Help assemble a symptom timeline, but leave judgement to a clinician or pharmacist. |
| Red: urgent or high risk | Do not wait for an AI response; severe, sudden or rapidly worsening symptoms require local emergency or health services. |
Prepare facts before asking
- When symptoms began and how they changed
- Current medicines and known allergies
- Existing diagnoses, recent tests and special circumstances
- The concern you most need addressed
Turn the answer into questions for care
Instead of asking “What disease do I have?”, ask AI to help formulate: “What should I tell the clinician?”, “Which warning signs require urgent action?” and “Which interactions should I ask a pharmacist about?” Then verify with an authoritative local health service.
AI may reduce the difficulty of describing a problem, but it cannot examine you, monitor deterioration or assume clinical responsibility. Decisions to start, stop or change treatment—or delay care—should never rest on a chat response alone.
References
Discover more from Geoffrey Chen
Subscribe to get the latest posts sent to your email.