• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
AI
Reaction

Lean 4's reported lack of a consistency proof and other hurdles for formal verification

A post says testing whether Lean code runs safely triggers Codex and Claude cybersecurity guardrails.

1 Source, 19h ago, first seen 19h ago

TLDR

A post sharing an article says Lean 4 has no consistency proof yet and describes β€œsea monsters” in its type theory as bugs that cannot be systematically squashed. The poster says checking whether Lean code runs safely triggers Codex and Claude cybersecurity guardrails, which they see as an opportunity for Mistral's Leanstral. They say the article's author considers these issues fixable, while the poster sees a difficult path to widespread formal verification.

Combined views

β€”

1 Source, first seen 19h ago

β€” likesβ€” commentsβ€” savesβ€” reposts

Combined views

β€”

1 Source, first seen 19h ago

β€” likesβ€” commentsβ€” savesβ€” reposts

Sentiment

Positiveβ€”β€”Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Sentiment

Positiveβ€”β€”Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

1 Source

Lewis Tunstall @ COLM πŸŒ‰@_lewtunThis is a fascinating article on some of the subtle and deep challenges one encounters with formal languages like Lean. I didn't realise that we don't yet have a proof of Lean4's consistency, and that there are some "sea monsters" concerning its type theory (basically bugs that you can't systematically squash) Another interesting observation is that testing whether Lean code runs safely triggers the cybersecurity guardrails of Codex / Claude. IMO this is a pretty good opportunity for @MistralAI's Leanstral :) As the author points out, all of these issues are fixable, but it does show that getting to a future where most software is formally verified won't be a simple task at all.19h
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI

    1 Source

    Lewis Tunstall @ COLM πŸŒ‰@_lewtunThis is a fascinating article on some of the subtle and deep challenges one encounters with formal languages like Lean. I didn't realise that we don't yet have a proof of Lean4's consistency, and that there are some "sea monsters" concerning its type theory (basically bugs that you can't systematically squash) Another interesting observation is that testing whether Lean code runs safely triggers the cybersecurity guardrails of Codex / Claude. IMO this is a pretty good opportunity for @MistralAI's Leanstral :) As the author points out, all of these issues are fixable, but it does show that getting to a future where most software is formally verified won't be a simple task at all.19h
    Today's Rank

    β€”

    Not ranked yet

    Today's Rank

    β€”

    Not ranked yet