AI Prompts for Decision Making: Think Clearer, Decide Faster

This page may contain affiliate links.

Most of us don't struggle to make decisions because we lack information. We struggle because we're carrying too much of it, all at once, in a head that's also trying to remember to defrost the chicken and reply to that email from Tuesday. AI won't make the call for you, and you shouldn't let it. But used properly, it's an outstanding thinking partner: it can hold the messy sprawl of a decision in front of you, ask the questions you've been avoiding, and strip an overcomplicated choice down to what actually matters. Here's how to use it well.

Stop asking AI what to do. Ask it how to think

The single biggest mistake people make is typing "should I take this job?" and hoping for a verdict. You'll get a bland list of pros and cons that helps nobody. The trick is to give the model a role, your context, and a job to do.

Start with something like this:

"Act as a sceptical but fair advisor. Here's my situation: [3-5 sentences of real context]. I'm deciding between [Option A] and [Option B]. Don't tell me what to do. Instead, ask me the five questions you'd want answered before you'd feel confident advising me, one at a time, and wait for my answer before asking the next."

That last instruction matters. Making the AI ask one question at a time forces you to actually think rather than skim. It also surfaces the information you were conveniently leaving out — the stuff that's usually the real crux.

Use prompts that fight your specific bias

Generic advice is cheap. What's useful is a prompt aimed at the particular way you're getting stuck. A few I come back to:

For overthinkers: "I've been going back and forth on this for weeks. Assume I already know the pros and cons. What's the smallest piece of new information that would actually change my mind? If there isn't one, say so plainly."

For sunk-cost thinking: "Here's a project I've put a lot into: [details]. Ignore everything I've already spent — money, time, effort. If I were starting fresh today with no history, would I begin this? Explain your reasoning."

For decisions that feel huge but aren't: "Rate this decision on a scale of 1-10 for how reversible it is, and how much it matters in five years. Be honest if I'm treating something trivial as life-changing."

For when you're deciding emotionally: "I'm angry/frustrated/scared about this. Separate my emotional reaction from the underlying facts, and tell me which parts of my reasoning are actually about the situation and which are about how I feel right now."

These work because they're narrow. A prompt that could apply to any decision usually helps with none.

Run a pre-mortem before you commit

This is the highest-value five minutes you'll spend on any significant decision. A pre-mortem assumes the decision has already gone wrong and works backwards to find out why.

"It's twelve months from now and the choice I'm about to make has turned out badly. Write the post-mortem. What were the three most likely reasons it failed? For each one, tell me what early warning sign I should watch for in the first month."

You can flip it too: "Now assume it went brilliantly. What did I do right that I might otherwise have skipped?" Between the two, you get a realistic picture of the risks and the things worth deliberately protecting.

If your decision involves buying kit — a laptop for a side project, a microphone for a podcast, a second monitor so you stop alt-tabbing like a maniac — it's worth running the pre-mortem on the purchase specifically. Ask the model what people typically regret about that category of product, then check the one-star reviews on Amazon UK to see whether the failure modes it predicted actually show up.

Get a second and third opinion from the same chat

One of AI's underrated tricks is that you can ask it to disagree with itself. After you've got an answer you like, push back:

"Now argue the opposite case as strongly as you can. What's the best version of the argument against what you just said?"

Then: "Given both sides, what would a cautious person do? What would an ambitious person do? What would someone who values their time over money do?"

Three framings, three different answers, and suddenly you can see which one you instinctively recoil from — which is usually a strong signal about what you actually want.

If you'd rather not build all these prompts from scratch, our AI Productivity Pack collects 90 ready-to-use ChatGPT and Claude prompts covering planning, email, meetings and focus (from £9). It's the done-for-you version of what's in this article, if you'd rather spend your thinking time on the decision itself than on writing the prompt.

Turn the decision into a written plan

A decision isn't finished until it's a set of actions. Once you've settled on a direction, close the loop:

"I've decided to [decision]. Turn this into a one-page plan: the first three actions I should take this week, the one thing I should stop doing, and the date I'll review whether this was right."

That review date is the bit people skip, and it's the bit that makes you better at deciding over time. Put it in your calendar. When it arrives, paste your original reasoning back into the chat and ask: "Here's what I predicted would happen. Here's what actually happened. Where was my thinking off?"

Do that a handful of times and you'll start noticing your own patterns — the decisions you rush, the ones you avoid, the risks you consistently over- or under-weight. That's the real payoff. The AI isn't getting smarter about your life; you are.

A quick word on what not to do

Don't paste in anything confidential — client names, salaries, personal health details — into a consumer AI tool unless you're comfortable with how that data is handled. Anonymise it: "a mid-sized UK retailer" works just as well as the actual name. And don't treat the output as verdict. Treat it as a well-read friend who's happy to be interrogated, has endless patience, and never gets bored of you going round in circles.

Used that way, AI genuinely does make you faster — not because it decides for you, but because it gets the fog out of the way so you can. Pick one decision you've been sitting on this week, open a chat, and try the five-questions prompt. You'll probably have your answer before your tea goes cold.