Which prompt to choose
A stuck problem and a hard decision are two different things. In the first case you don't see the way out; in the second you see them all and can't choose. The third prompt is for when the problem is your head, not the situation: you're going in circles. Choose the right one.
- You're stuck and don't see any solutions: first prompt, which opens up options you hadn't considered.
- You have several options and can't choose: second prompt, which weighs them against your own criteria.
- You're ruminating without concluding: third prompt, which stops the spinning.
How to do it
- Open the AI assistant.
- Describe the situation honestly, including the uncomfortable details: the AI works only on what you give it, and if you leave out the real crux it gives you advice for the wrong problem.
- Copy the prompt and send.
- Don't stop at the first answer: go again with "give me the option I wouldn't dare consider," "what's the risk I'm ignoring?"
The operational syntax for unsticking a problem:
I'm stuck on this situation: "[describe the problem with the real details, even the uncomfortable ones]."
Don't give me a solution right away. First:
1. Help me understand what the real problem is beneath the one I described.
2. List 5 possible routes, including 2 that most people wouldn't consider.
3. For the 2 most promising, tell me what it would take to make them work.
Ask me the questions you need to frame it better, instead of guessing.
The operational syntax for deciding between several options:
I have to decide between these options: "[list the options]."
The decision is about: "[context and what's at stake for me]."
Help me like this:
1. Ask me what the 3-4 criteria are that really matter TO ME (not generic ones).
2. Build a table that weighs each option against those criteria.
3. Tell me which is the most solid choice and what risk I carry by choosing it.
4. Then tell me when, on the contrary, it would be worth choosing differently.
Don't just go along with me: if one of my options is weak, tell me.
The operational syntax for stopping the rumination:
I've been ruminating for days on this without concluding: "[the situation]."
Help me get out of the spinning:
1. Distinguish what in all this depends on me and what I can't control.
2. On what depends on me, what's the single step I could take by tomorrow.
3. On what I don't control, give me a realistic sentence to stop torturing myself over it.
Be direct and concrete, no generic consoling phrases.
After the answer, always ask for the opposite angle: "what's the strongest reason to do the opposite of what you advise me?" A decision that holds up even against the opposing argument is a robust decision.
A real example
Giulia didn't know whether to leave a safe but dull job for a riskier offer. She'd been going in circles for weeks. She used the deciding prompt: the AI asked her for her real criteria, and the true ones emerged, not the textbook ones: growth, her family's peace of mind, future regret.
In the table the "I stay" option lost on growth and regret, won on immediate peace of mind. Giulia asked for the opposing argument, and the AI reminded her that the perceived security of the old job was itself a risk (the company was losing market share). It didn't tell her what to do. It gave her a clean map of a choice that before was only anxiety. She accepted the offer.
When it does NOT work (and how to fix it)
If the AI always agrees with you
It's the most insidious flaw when you ask for advice: models tend to go along with you, a tendency the technical people call sycophancy. If you present the problem in a slanted way, you get confirmations of your idea. The remedy is in the prompt: "don't just go along with me, challenge my premises if they're weak," and always ask for the opposing argument.
If it gives you textbook advice, disconnected from your life
That means you gave it little context. A generic answer comes from a generic question. Add the real constraints: the money you have, the people involved, what you're not willing to sacrifice. The more concrete the situation, the less the advice smells of a horoscope.
If the decision is too important to entrust to an AI
You're right, and it has to be said clearly: on serious medical, legal, or financial choices the AI helps you frame the questions, not deliver the verdict. Use it to arrive prepared at the doctor, the lawyer, or the consultant, with the right questions already ready. The professional decides with you; the AI gets you there less confused.
A tip from someone who actually uses it
When a decision unsettles you, first write to the AI what you instinctively want to do, then ask it to dismantle precisely that choice. Seeing your impulse put to the test from an outside, non-emotional point of view is what helps you understand whether it's a decision or just an escape. The AI's distance from your feelings, here, is an advantage.
Frequently asked questions
Can the AI help me with personal problems too, not just work ones?
For putting order in your thoughts and seeing a situation from the outside, yes, and many use it that way. But on real suffering, relationships in crisis, or psychological distress it doesn't replace a person: neither a professional nor a friend. Treat it as a sheet of paper to think out loud on, not as someone who understands you.
How do I avoid delegating my decisions entirely?
Use the AI for the analysis phase (causes, options, consequences) and keep the choice phase for yourself. A practical rule: the AI can tell you what to consider, never what to want. What you want only you know, and no table calculates it.
Are decisions made with AI better than gut ones?
Not always, and it's the misunderstanding to dispel. Instinct, on topics you know well, is often right and should be listened to. The AI doesn't replace it: it puts it to the test. It serves to stop you mistaking fear for prudence and enthusiasm for certainty. On a field you've mastered, the analysis confirms the instinct or corrects it; it doesn't cancel it.