Most important decisions are not made with bad information. They are made with incomplete information, biased mental frames, and time pressure. AI does not solve all three problems, but it can meaningfully reduce the first two — if you use it as an interlocutor before deciding, not as a validator after.
Why we make poor decisions
The problem with decisions is not a lack of intelligence. The human brain uses cognitive shortcuts that worked well in simple environments and fail in complex ones. We confirm what we already believe, we overweight recent events, we ignore second-order consequences, and we mistake familiarity for correctness.
Add to this that important decisions are usually made under pressure: deadlines, emotions, other people’s expectations. That context amplifies biases rather than compensating for them.
The result is that we tend to decide in the direction we already wanted to go, selecting arguments to justify it afterward. We rationalise more than we reason.
Where AI fits in
AI does not have access to your real situation, does not know your values, and cannot predict the future. What it can do is expand the space of consideration before you close the decision.
There are three concrete uses where this adds the most value.
Identifying hidden assumptions. When you describe a situation and your tentative decision, AI can point out the things you are taking for granted without having examined them. Its observations are not always correct, but the exercise of evaluating whether they are already clarifies your thinking.
Generating alternatives you had not considered. The biggest risk in many decisions is seeing only two options when there are four. Asking AI to propose alternatives to the ones you have framed does not guarantee they will be better, but it widens the menu.
Exploring the opposite case. If you already have a position, asking AI to build the strongest argument against it is a fast way to find the weak points in your reasoning — not to change your mind by default, but to reinforce the decision or to discover that you need more information first.
A simple flow for complex decisions
The process I find most useful has three steps and takes no more than twenty minutes.
First, I write the decision as I see it: what I am considering, why I lean toward one option, and what is making me hesitate. Writing it down usually clarifies things before AI enters the picture.
Second, I pass that text to AI with this prompt: “What assumptions am I taking for granted? What am I ignoring? What alternatives have I not considered?” The response is not the decision. It is material to think with.
Third, I ask for the strongest argument against my preferred option. Not to be automatically persuaded, but to see whether I have an answer to that argument. If I do not, I need more information. If I do, I decide with more confidence.
AI does not tell you what to decide. It tells you what you have not looked at yet.
What AI cannot decide for you
AI does not know your values, your relationships, or what actually matters in your life. It can give you a structured analysis, but it cannot weigh that against what you feel is worth it.
Decisions with deep personal consequences — career changes, relationships, long-term commitments — have dimensions that more information does not resolve. They are resolved by clarity about who you are and what you want, and that is not something you can outsource.
Using AI as a copilot works precisely because you do not hand over the wheel. You ask it to point out your blind spots. But the one who decides — and who lives with the decision — is still you.
