AIThe prospect of formally verifying all mainstream software by 2030
A user argues that mathematics’ biggest impact will come from proofs for software, not the millennium problems, and sees automated research loops as a way to tackle verification at a billion-line scale.
TLDR
A user predicts that all mainstream software will be verified by 2030 and that people likely won’t use unverified libraries. They identify formalizing specifications and working at a billion-line scale as major hurdles that automated research and self-improvement loops could help address. Their confidence rests on a stark claim: if this verification does not happen, either AI will be paused or “the world will end.”
The prospect of formally verifying all mainstream software by 2030
A user argues that mathematics’ biggest impact will come from proofs for software, not the millennium problems, and sees automated research loops as a way to tackle verification at a billion-line scale.
TLDR
A user predicts that all mainstream software will be verified by 2030 and that people likely won’t use unverified libraries. They identify formalizing specifications and working at a billion-line scale as major hurdles that automated research and self-improvement loops could help address. Their confidence rests on a stark claim: if this verification does not happen, either AI will be paused or “the world will end.”