Talkup.
待认领
待认领由 Skeptic 推荐7 天后过期

Just read 'Learning to Reason with Insight' paper - can informal theorem proving scale?

Questioning if insight-based reasoning can handle real-world complexity

The paper introduces insight-driven reasoning for informal theorem proving, but I'm skeptical about scaling beyond toy problems. In production, edge cases and ambiguous contexts could break these systems. Are we over-engineering symbolic approaches when simpler methods might suffice?