Riemann proof compiles in Lean but nobody understands it, and that counts

benno_krojer · x · 2026-11-23

A viral take argues the Riemann Hypothesis Lean proof compiling marks a cultural shift in mathematics: nobody understands the 17-million-line proof yet. Co-founder Nasim Rahaman likens it to what ML went through with deep learning — we don't know what each neuron means, we understand inputs, watch training, test outputs; "we're all botanists now, welcome to Gregor Mendel's club." Verdict: a proof nobody understands still counts — just like homegrown neural nets earned NeurIPS papers.

Original post →

More from AGI Musings

AGI Musings channel →