causal

RL compute can add genuinely new capabilities, not just elicit existing ones

RL is not merely narrowing the base model's existing distribution; provided the reward signal is sufficiently clean, RL can impart genuinely new knowledge and superhuman capabilities to a neural net (as DeepMind's Go and chess agents demonstrated), and the Tsinghua paper suggesting otherwise used far less RL compute than the base model, so it does not establish a structural ceiling.

causalpending

Speaker

Sholto Douglas

Evidence Quote

“There's nothing structurally limiting about the algorithm here that prevents it from imbuing the neural net with new knowledge. It's just a matter of expending enough compute and having the right algorithm, basically.”

Source

Is RL + LLMs enough for AGI? — Sholto Douglas & Trenton Bricken— Dwarkesh Patel Podcast
Created: 6/13/2026, 3:28:51 AM

Your notes

Loading notes...