Why I Stopped Using AI for Brainstorming
AI brainstorming feels productive and produces nothing I'd actually build. The comfortable answers were the tell.
I used to open a chat and ask for 'ten startup ideas in X.' It felt like progress: a full page of options in seconds. Then I noticed I never built any of them. The list was plausible, safe, and completely devoid of the weird conviction that actually gets something shipped.
The problem is the model optimizes for agreeable, common answers. Real brainstorm energy comes from a bad idea you're slightly obsessed with, and no prompt extracts that. The AI's 'ten ideas' are a committee's middle, and committees don't start companies or write the essay only you could write.
I still use AI to pressure-test an idea I already have, to find the weak spot, to argue the other side. That's where it's honest help. But generating the seed? That stays a pen and a blank page, because the discomfort of staring at nothing is the only reliable source of something original.
If your brainstorm output is always polished and never risky, you're not brainstorming, you're summarizing the average. Close the tab and sit with the silence for ten minutes. The idea that survives the silence is the one worth building.