Large language models appear intelligent in only narrow domains (mathematics, coding, law) where reasoning is substrate-based (symbol manipulation itself aids thought) and where answers can be self-corrected (code can be tested, proofs verified).

causalpending

Speaker

Yann LeCun

Evidence Quote

LLMs exhibit intelligence only in domains where the reasoning is substrate-based (symbol manipulation aids thought) and where answers can be self-corrected... they basically do retrieval... extremely good at regurgitating knowledge at the right time.

Source

Yann LeCun: Special Lecture on AI and World ModelsAl-Khwarizmi Applied Mathematics Webinar
Created: 8/12/2026, 10:03:11 PM

My Notes

Loading notes...