Lee Cronin: LLMs are not intrinsically intelligent, just echoes of their training data

AlexTensor · x · 2026-09-13

Chemist Lee Cronin argues that conversing with an LLM means interacting with "the intelligence that created the training data," not the model itself. He acknowledges hidden structures in language shaped by cultural evolution but contends the silicon-instantiated models are not intrinsically intelligent.

Original post →

More from AGI Musings

AGI Musings channel →