We have proof automation now | Hacker News

A Hacker News discussion argues that LLMs, paired with proof irrelevance, could automate many formal proofs and make dependent type systems more practical.

Commenters note that automation still depends on good problem framing, proof strategy, and proof engineering, not just model output.

The thread also highlights AWS’s LNSym AArch64 semantics and simulator as a way to check optimized assembly against Lean implementations.

If these tools work well together, they could lower the cost of formal methods and improve confidence in low-level verified code.