Every AI utopia starts by removing the one variable that doesn't fit the model. The human.

I've been rereading Hayek's The Fatal Conceit.
It's a short book with one argument and right now it reads like a direct response to half the AI pitches I see.
The temptation is always there: engineer society like a machine. Describe the world in equations. Eliminate irrationality. Build the perfect system.
It makes sense if you come from physics or engineering. If you can split the atom and describe it with math, the logic follows: why not society? The universe has laws. Surely people do too.
That's the logic behind planned economies. Same pitch, different branding, behind AI-governed cities.
Hayek's problem with this isn't the math. It's the model of the human. People run on tradition, local context, informal rules, contradictions that never fully resolve. Not a bug. Just what a person is.
Knowledge in any society is dispersed. It lives in prices, in local decisions, in signals no central planner can collect in full. Markets aren't an ideology in Hayek's framing — they're the only mechanism that lets this knowledge move. Centralize, and you destroy exactly what you were trying to optimize.
I keep thinking about this when someone pitches AI-managed monetary policy, or a financial system where the human is an input variable.
The technology isn't the problem. What these systems discard first is the person who doesn't fit the model. That's where the information was.