matio

Electronics and embedded-systems engineer and programmer. I came to AI alignment from an economic angle, on the working hypothesis that the core of the problem is one of economic incentives rather than a purely technical control problem and I’d like to be shown where that hypothesis breaks.

Gradient exhaustion is my attempt to write down a minimal, falsifiable model of economic disempowerment (humans priced out of the economic loop by compounding AI capability) and to have it broken rather than admired. Closed-form where possible, verified several independent ways, honest about the assumptions doing the work. Code and derivations: github.com/​​MatioRIGARD/​​gradient-exhaustion

Outsider to economics; based in Southeast Asia. Corrections welcome, especially the uncomfortable ones.