Stop asking AI for answers. Ask what you are missing

Most people use these tools like a search engine that talks. Question in, answer out. That works for converting a PDF. It works badly for anything that matters, like choosing between two job offers or working out why a side project has no users, because for those your question is usually the problem. You do not know what you do not know, so you ask the wrong thing and get the wrong thing answered very confidently.

So flip it. Before asking for an answer, ask what you are missing.

The basic version: I am trying to do X. Here is my situation, in two to five sentences. Before you give me any advice, what questions should I be asking that I am not, and what information would change your answer?

Nothing clever about it. But instead of a generic list of tips you get things like: is the second offer remote, does your current employer match the pension, have you checked whether the higher salary moves you into a different tax position. Suddenly you are thinking about the decision rather than reading a listicle.

Why it beats hunting for a perfect prompt: these models are good at knowing what is usually relevant to a type of problem, because they have seen millions of versions of "should I take this job". What they do not know is your version. A normal question makes the model guess the blanks with the most average assumption available. Asking what you are missing makes it show you the blanks so you can fill them in yourself.

It also sidesteps a known failure. These things tend to go along with how you framed the question, so a bad assumption baked into your question comes back baked into the answer. Asking for the missing questions drags the assumption into the open.

Five shapes that work:

Big decisions: I am deciding between A and B, what are the questions people in my position usually regret not asking. The word regret is doing the work there, it pushes the answer toward what people learn the hard way.

Work projects: here is my plan, what would a senior person in this field immediately ask that I have not addressed. Useful before the meeting where somebody does exactly that.

Medical appointments: I have an appointment about this symptom, what should I ask and what should I track beforehand. Not asking for a diagnosis, using it to get more out of fifteen minutes with an actual doctor, which is a genuinely good use.

Money: I am about to sign a lease or take a loan, what terms and fees do people commonly overlook. It will not know your contract. It gives you a checklist to read the contract with.

Learning: I want to learn this, what do beginners usually not realise they need first. Saves weeks, because most beginner frustration is a missing foundation nobody mentioned.

Then the second level, once that feels natural: ask me questions one at a time to understand my situation, and after five questions give me your advice. Slower, and the closest thing to talking to a smart friend who is actually listening.

One warning. Better questions do not make the answers correct. These tools are still wrong, outdated and occasionally confidently invented, particularly about law, prices and anything that changed recently. Use it to find the right questions and then check the important answers against a real source or a real person. Let it help you think rather than think for you.