The analogy
Think of the leaflet in the doctor's waiting room. It gives you useful, general information — how to sleep better, what hypertension is, why moving is good for you — but it knows nothing about you in particular. It's a starting point, not a visit. And think of the nurse friend you run into at dinner: they give you a common-sense pointer and, if you describe something that doesn't sit right with them, they tell you "go get it checked."
The AI on health sits exactly there: general information and common sense, never the visit. Useful for understanding and arriving prepared, never for replacing the one who really examines you.
How it really works
On health and wellbeing the AI keeps an educational stance: general information, habits, prevention, explanation of terms. It doesn't make a personalized diagnosis and, when warning signs emerge, it refers you to the doctor. The point is that it doesn't see your body, doesn't know your medical history, doesn't have your test results: it works on what you write to it and on what it has read in general. Worth flagging is a recent change: caution notices appear far less than they used to in health answers, so prudence is increasingly on your shoulders, not automatic.
What you can do in practice
- Understand in simple words a term, a medical report in broad strokes, how a certain therapy works.
- Set up healthy habits: sleep rhythm, movement, hydration, meal organization.
- Prepare for the visit: write the timeline of symptoms, the questions to ask, the drugs you take, so you don't forget anything at the doctor's.
- Get a rough idea of the urgency ("is this an emergency-room thing?"), without taking it as gospel. In the face of a serious signal, contact a doctor or emergency services right away.
A common misconception
People think that, if you describe the symptoms well, the AI tells you what you have. It can list generic possibilities, but it doesn't examine you and doesn't see the tests, and above all symptoms resemble each other between a trivial thing and a serious one: the same headache can be a thousand things. Self-diagnosis by chat is risky precisely because it seems precise while it's guessing on incomplete information. A list of hypotheses isn't a diagnosis.
Frequently asked questions
Can I ask it about my symptoms?
To understand in general what it's about and get an idea, yes. To know what you actually have, no: that requires a visit. Use what it tells you to arrive at the doctor with the right questions, not to decide on your own that it's nothing or that it's extremely serious.
Does it help me stay healthy?
Yes, and here it's at its best: habits, prevention, organization, clear explanations. On sleep, movement and nutrition it's an excellent reminder and a good educator, because those are territories of general common sense and not of diagnosis.
And if it's something urgent?
Don't waste time with the chat. In the face of serious symptoms or ones that worsen quickly, contact a doctor or the emergency services. The AI isn't equipped to handle an emergency and every minute spent questioning it is a minute taken away from those who can really help you.
Many people now use it as a doctor, so can it be trusted?
No: widespread doesn't mean safe. The AI hasn't seen you, shows fewer and fewer caution notices and can seem convincing precisely while it's wrong, because it speaks with a confident tone even when it's guessing. For the body you need someone who really examines it, with the tools and the responsibility to do so. Using it as a doctor is convenient until it's the time when the error counts.