How to use AI to
think through a decision

1. Name the choice and what you want from it

Start with a specific decision. “What should I do with my free time?” leaves a lot for the AI to guess. “Should I take an evening photography course or learn on my own?” gives the conversation a useful starting point.

Then explain what you want the decision to achieve.

For our example, imagine you want to get comfortable using your camera. You are choosing between an eight-week class and self-study. You work full-time, often feel tired after work, and have previously abandoned tutorials.

This is an illustrative scenario. The prompts below are examples you can adapt.

I’m deciding between an eight-week evening photography class and learning on my own. I want to understand my camera and practise regularly. I work full-time and often abandon online tutorials. Help me clarify the decision. Ask me one question at a time before comparing the options.

That last instruction gives you room to think. You can also say what kind of help you need: exploring possibilities, challenging an assumption, or organizing thoughts you already have.

2. Separate requirements from preferences

Some considerations determine whether an option is possible. Others make it more appealing.

A class within your budget might still be unsuitable if you cannot get there on time. A flexible self-study plan might fit your calendar but offer less of the structure you want.

Help me separate my requirements from my preferences. I need to keep Thursday evenings free and stay within my budget. I would like feedback on my photos and a reason to practise each week. Ask about any important constraints I have missed.

Be specific about vague words. “Affordable” needs a limit. “Convenient” might mean a short journey, a predictable schedule, or being able to skip a week.

Choose your three most important considerations. For this example, they could be regular practice, useful feedback, and a schedule you can keep. Those priorities make the comparison more meaningful than a long list where every point appears equally important.

3. Sort facts from assumptions

AI can turn incomplete information into a confident-sounding answer. Before asking for a recommendation, make the gaps visible.

Separate what I have told you into facts, assumptions, and unanswered questions. Don’t fill in missing details. Which unanswered question could most change the decision?

Suppose the course description lists Wednesday classes and practical assignments. Your notes might look like this:

The final row matters. “Self-study never works for me” is a broad conclusion. Your earlier attempt may have lacked a clear project or time set aside for it.

Ask the AI what evidence would help, then get that evidence from the appropriate place. Check the course’s current description or contact the organizer. A chatbot’s guess about the timetable, cancellation policy, or tutor cannot settle those questions.

4. Compare the options against your real week

Now ask for a comparison using the priorities you identified.

Compare the class and self-study against regular practice, useful feedback, and a schedule I can keep. Use only the information I have provided. Explain the trade-offs in plain language and mark anything uncertain.

A comparison becomes more useful when it includes the ordinary details of living with each choice.

For the class, that could mean travelling after work, eating dinner later, and keeping Wednesday free for eight weeks. For self-study, it could mean choosing a project and protecting two practice sessions each week.

Describe what an ordinary week would involve for each option. What would I need to arrange, and what might make me skip it?

Be cautious with numerical ratings. A table that gives the class 8.7 and self-study 7.9 may look precise, but the numbers can simply reflect guesses. If you use scores, choose the criteria and weights yourself, and check whether small changes reverse the result. A short explanation of the trade-offs may tell you more.

5. Challenge the option you already prefer

You may notice that you are hoping the AI recommends a particular option. That preference is useful information. Give it some scrutiny.

I’m leaning toward the class because I want structure. What is the strongest reasonable case against it, based on what I’ve told you? Which concern should I investigate before enrolling?

Then examine the alternative fairly:

What would need to change for self-study to become a realistic option? Suggest one specific change, keeping my time and budget constraints in mind.

Perhaps self-study would work better with a small project and a friend who also wants to practise. Perhaps that still would not provide the feedback you need.

The point is to understand why an option fits. AI can miss context or follow the direction of your question too readily, so correct it when it introduces an assumption you do not share.

6. Choose a useful next step

Sometimes the next step is making the decision. Sometimes it is resolving one uncertainty first.

In our example, you might ask the tutor about feedback, try the journey at the relevant time, or test two scheduled practice sessions at home.

What is the smallest practical step I can take this week to resolve the most important uncertainty? Explain what I could learn from it and what it would leave unanswered.

Choose an action that could actually change your view. Another generic comparison is unlikely to help if the real question is whether you can attend the class reliably.

Finish with a short decision note:

Summarize my current preference, the reasons behind it, the main uncertainty, and my next action. Include what new information would make me reconsider.

Give yourself a stopping point, too: “I’ll decide on Friday after the tutor replies.” You can revisit the note when you have new information, rather than restarting the whole conversation whenever you feel unsure.

A prompt to get started

You do not need an elaborate prompt. A little honest context is enough:

I’m deciding between [option A] and [option B]. I want [goal]. My main constraints are [time, budget, commitments, or other limits]. I’m leaning toward [option] because [reason], but I’m unsure about [question]. Ask me one question at a time, separate facts from assumptions, and help me identify a useful next step.

For another practical example, see our guide to talking things through and preparing for an interview.

A useful conversation ends with something you can check or do. You do not need to remove every doubt before taking that next step.