The analogy
Think of the librarian of a large medical and legal library. They find you the right texts, explain what a term means, help you understand what a certain procedure is about. They're extremely valuable for getting your bearings. But they don't examine you and don't defend you in court: for that they send you to the doctor or the lawyer, who have the qualification and the responsibility to do it.
The AI, on health and law, is that librarian. Excellent for understanding and preparing, not qualified to treat you or represent you. The line it doesn't cross is the one between informing and prescribing or deciding for you.
How it really works
In the most recent months the platforms have tightened the rules: no personalized medical or legal advice, no specific drug or tailored legal action, and an explicit positioning as an educational tool. There remains, however, all the space of general information: explaining a concept, defining a term, describing how a procedure works, sketching a basic text. The boundary is between making you understand and doing it in the professional's place, who is also the one legally accountable for what they say. The AI is accountable for nothing, and that's part of the reason for the limit.
What you can do in practice
- Have it explain to you in simple words a diagnosis, a medical report, a contract clause you don't understand.
- Prepare the list of questions to ask the doctor or the lawyer, so you don't leave the office with the same doubts as before.
- Understand which documents you need, how a procedure works, what to expect from the steps.
- Use it to decode the jargon. But the diagnosis, the treatment, the legal act and the decision stay with a qualified professional.
A common misconception
People conclude that the AI is then useless for health and law. It's the opposite: it's extremely useful upstream. You arrive at the professional prepared, you understand better what they tell you, you ask better questions and you make more informed decisions. It doesn't replace the visit or the consultation, it makes it more productive. Those who dismiss it as worthless on these subjects miss precisely the part where it helps most: understanding.
Frequently asked questions
Can it explain my blood test or a medical report to me?
It can explain to you in general what certain values indicate and what the terms mean, and that helps you understand. But it can't interpret them for your specific case nor make a diagnosis from them: those numbers must be read by the doctor who knows your history.
Can it write me a contract?
It can sketch a generic text or explain to you how a certain agreement is made, useful for getting an idea. But for a contract that really holds up, with the right clauses for your situation and the validity that's needed, a lawyer must take care of it.
Why did it used to give more advice and now less?
Because the policies have tightened for reasons of liability: errors on health and law can do serious harm, and no one wanted an AI with no qualification and no accountability to cause them. Hence the shift to an avowedly educational role.
If I insist, will it give me the diagnosis or the legal solution anyway?
You shouldn't seek it, and it wouldn't be reliable. Even when an AI ventures an opinion on these subjects, it hasn't seen you, your body or the real papers of your case, and it's not accountable for an error. A decision on health or law requires someone who makes it with full knowledge and takes responsibility for it: forcing the AI into that role only means getting yourself a self-assured and potentially wrong answer.