Every experienced PM has a story about a decision that looked wrong on paper but turned out right. And another about a decision that looked perfect in the spreadsheet and failed spectacularly.
That tension — between data and judgment — is the core of the job. AI changes the inputs. It doesn’t change the tension.
The Trap: More Data, Same Decisions
When PMs first get access to powerful AI tools, the instinct is to use them for analysis. Run more queries. Synthesize more feedback. Generate more scenarios. And suddenly you have ten times the data to support… the same decision you would have made anyway.
This is the confirmation bias problem at scale.
AI can generate a compelling narrative for almost any position. If you’re not deliberate about how you use it, you’ll use it to make yourself feel more confident about decisions you’ve already made subconsciously.
Using AI to Challenge, Not Confirm
The better approach: use AI adversarially.
Before finalizing a major decision, explicitly ask your AI tool to argue the opposite position. Ask it to find the three most compelling reasons your strategy will fail. Ask it to role-play as your most skeptical board member or your most frustrated customer.
You’re not looking for it to be right. You’re looking for it to surface the objections your own motivated reasoning is suppressing.
Where AI Genuinely Expands PM Judgment
Competitive landscape scanning. AI can maintain a continuously updated view of market movements, competitor releases, and customer sentiment across sources no team could manually track.
Scenario modeling. What happens to retention if we change pricing? If we ship this feature in Q2 instead of Q3? AI can run variations that used to require dedicated analyst time.
Customer signal synthesis. Connecting the dots between support tickets, NPS verbatims, app store reviews, and social mentions — at scale, in real time — turns weak signals into conviction.
The Irreplaceable Part
None of this replaces the thing that makes great PMs great: the ability to sit with ambiguity and make a call.
AI gives you better inputs. The decision is still yours.