How to Use AI to Make Better Decisions

Most people think AI is for writing stuff. Emails, essays, social media captions. And it's fine at those things.

But the most valuable thing I use AI for has nothing to do with writing. I use it to think.

Specifically, I use it to make better decisions. Career decisions, money decisions, "should I move across the country" decisions. The kind of stuff you normally talk through with a friend at 1am, except the friend never gets tired and has read basically everything.

Here's why this works so well, and how to actually do it.

Why We're Bad at Decisions

Humans are terrible at making decisions. Not because we're dumb, but because our brains are wired with dozens of cognitive biases that distort our thinking.

You anchor on the first number you hear. You overweight recent experiences. You avoid losses more than you pursue gains. You confuse "familiar" with "good." You rationalize whatever you already want to do.

We know all this from behavioral economics. Kahneman won a Nobel Prize for documenting it. But knowing about biases doesn't fix them. You can't just decide to stop being biased. That's like deciding to stop having an accent.

What you can do is talk to something that doesn't share your biases.

AI as Devil's Advocate

The simplest way to use AI for decisions is to tell it what you're thinking and ask it to argue against you.

Try this. Open ChatGPT or Claude and type something like:

"I'm thinking about quitting my job to go freelance. Here's my situation: [describe it]. I'm leaning toward doing it. Please argue against this decision as strongly as you can. Be specific about risks I might be ignoring."

What you'll get back isn't just generic "well, it's risky" advice. It'll point out specific things: your emergency fund covers 3 months but freelance income takes 6-9 months to stabilize, you'd lose employer health insurance in a state where individual plans cost $600/month, your industry has seasonal slowdowns you might not be accounting for.

This isn't the AI making the decision for you. It's forcing you to confront the parts of the picture you're conveniently ignoring.

I do this with almost every major decision now. The key is asking it to argue against whatever you're already leaning toward. Because that's the side you're blind to.

The Pre-Mortem Technique

There's a strategy from psychology called a "pre-mortem." Instead of asking "will this work?" you imagine it's a year from now and it failed, then ask "why did it fail?"

AI is perfect for this.

"Imagine it's March 2027 and I bought this rental property and it was a disaster. What went wrong? Be creative and specific. Consider market conditions, maintenance costs, tenant issues, opportunity cost, and anything else."

The AI will generate 8-10 plausible failure scenarios you probably haven't thought about. Some will be obvious. But usually 2-3 will make you go "oh, I hadn't considered that at all."

That's the value. Not that the AI knows the future. It doesn't. But it can systematically explore the space of possibilities faster than you can, without the emotional attachment that makes you skip over the scary scenarios.

Mapping Out What You Actually Want

Sometimes the problem isn't evaluating options. It's figuring out what you actually want.

This sounds simple but it's not. People are shockingly bad at knowing their own preferences. You say you want a high-paying job but what you actually want is flexibility. You say you want to live in New York but what you actually want is to feel like things are happening around you.

AI is weirdly good at helping you dig into this. Try:

"I'm trying to decide where to live next. I've been considering Austin, Denver, and staying in LA. But I think the real issue is I'm not clear on what I'm optimizing for. Ask me a series of questions to help me figure out what actually matters to me. Go one question at a time."

Then just have a conversation. The AI will ask things like "When you imagine your ideal Tuesday, what does it look like?" or "What did you like most about the last place you lived?" or "If money were irrelevant, would your list change?"

After 10-15 exchanges, you'll have a much clearer picture of your actual priorities. Not the ones you think you should have, but the ones you actually have.

Stress-Testing Your Logic

One of my favorite prompts:

"Here's my reasoning for [decision]. Find the logical holes. Where am I making assumptions? Where is my evidence weak? What am I conflating?"

This is like having a smart, patient editor for your thinking. The AI will identify places where you're treating correlation as causation, where you're generalizing from one example, where your conclusion doesn't actually follow from your premises.

You don't have to agree with everything it says. But going through the exercise of defending your reasoning against specific objections makes your thinking sharper.

The "10 People" Framework

Here's a technique I stumbled into that works really well:

"I'm deciding whether to [thing]. Simulate the advice I'd get from 10 different people: a risk-averse accountant, a startup founder, my mom, a therapist, a 70-year-old looking back on life, a financial advisor, my best friend who always says yes, a pessimist, an optimist, and a philosopher. Give me each person's honest take in 2-3 sentences."

What this does is show you the decision from radically different value systems. The accountant cares about numbers. The therapist cares about your emotional state. The 70-year-old cares about regret.

No single perspective is "right." But seeing all of them at once helps you understand which perspective resonates most with you, which tells you something important about what you actually value.

What AI Won't Do

I want to be honest about the limits here.

AI won't make the decision for you. It shouldn't. These are your life choices and you have context that no model can fully capture. Your gut feeling matters. Your relationships matter. The stuff that's hard to put into words matters.

AI also doesn't know the future. It can help you think through scenarios but it can't tell you if the housing market will crash or if your startup idea will work. Anyone who tells you AI can predict that is selling something.

And it can sometimes be too agreeable. If you push back, most AI models will cave and tell you what you want to hear. Be aware of this. The best approach is to explicitly tell it not to be nice: "Don't be diplomatic. Be brutally honest. I need the hard truth, not reassurance."

Start Small

You don't have to start with life-changing decisions. Try it with something low-stakes first.

"I'm trying to decide between these two laptops. Here are the specs and my use case. Which would you pick and why? Then argue for the other one."

Or: "I can't decide if I should take this online course. Here's what it costs, what it covers, and what I'm hoping to get out of it. Is this a good use of $500?"

Once you see how useful it is for small stuff, you'll naturally start using it for bigger things.

The whole point isn't that AI is smarter than you. It's that AI thinks differently than you. And when you're stuck in your own head, what you need most is a different perspective.

Your brain is great at many things. Seeing its own blind spots isn't one of them. Now you have a tool that can help with that.

If you want to go deeper on using AI for real life stuff, check out readlaboratories.com/learn or shoot me an email at jake@readlaboratories.com. I'm always happy to help people figure this stuff out.