The challenge with AI safety and formal verification is inherent to computation itself: as soon as you want computation to be genuinely useful and Turing complete, you must accept Gödel's incompleteness and computational irreducibility, which means you cannot box in what the system will do in advance—there is a fundamental tradeoff between computational power and provable safety

causalpending

Speaker

Stephen Wolfram

Evidence Quote

“if you want to say I want to know what happens you're boxing yourself in”

Source

Stephen Wolfram: Complexity and the Fabric of Reality | Lex Fridman Podcast #234— Lex Fridman
Created: 8/11/2026, 1:35:57 AM

Your notes

Loading notes...