Richard Sutton – Father of RL thinks LLMs are a dead end

What people said

solumos · 2026-07 · Reference

Google's definition is sufficient[0]. I'd also go so far as to say that LLMs aren't even AI, but that can be a bit pedantic. I align with Richard Sutton's reasoning, in that LLMs (and GPT models more generally) aren't "intelligent" in the same way that RL models are[1]. [0] - [LINK] [1] - [VIDEO]

Read the original HN comment ↗

mbesto · 2026-07 · Reference

This is basically the entire argument by Richard Sutton happening in real time: [VIDEO] Right now, we keep creating new models with larger and larger weights and throwing hardware at it. His argument is that this is a dead end and ultimately we'll eventually go back to purpose fit algorithms like we've always done in the history of AI development.

Read the original HN comment ↗

kbrkbr · 2026-05 · Reference

I think they are not actually intelligent. Fix all random seeds and other sources of randomness, and try the same prompt twice, and check how intelligent that looks, as a first approximation. On a more technical level very serious people have voiced doubts, for example Richard Sutton in an interview with Dwarkash Patel [1]. [1] [VIDEO]

Read the original HN comment ↗