Same model, different wiring
If everyone gets roughly the same intelligence, people and companies stop differing by what they can think, and start differing by what they plug it into.
everyone got the same socket
trying to answer →Why do AI products converge?Which parts of an AI startup are actually defensible?
When factories electrified, every one of them got the same current from the same grid. For decades productivity barely moved. The gains came only when owners stopped swapping a big motor for the old steam engine and redesigned the floor around small motors at every machine.The economist Paul David made this argument in 1990, in a paper about why computers weren’t showing up in the productivity numbers either. Same power, different wiring, very different results.
Frontier models look headed for the same shape. Capabilities converge; the gap between the best model and the fifth-best keeps closing. So the intelligence layer stops being where people differ. What’s left is what each of us connects it to: our information, experience, networks, taste, incentives and the institution we sit inside.
That applies to people as much as companies. Two analysts with the same model are only as different as the questions they bring and the context they can feed it. Uncomfortable corollary: a person with no distinctive context is, to the model, interchangeable with everyone else using it.
Open: does this make differences between people bigger (context compounds) or smaller (the floor rises)? I suspect bigger at the top and smaller in the middle, but I haven’t earned that yet.