causal

RL requires prior probability on correct solutions

For RL-based training to work, the model must already have some nonzero prior probability of producing the correct solution, which is why mind-stretching amounts of human expert trajectories are needed in every skill the model should master.

causalpending

Speaker

Dwarkesh Patel

Evidence Quote

“the model must have at least prior some probability to anticipate the correct solution”

Source

Dwarkesh Patel Podcast - The data black hole at the center of AI— Dwarkesh Patel Podcast
Created: 11/17/2026, 6:11:50 AM

Your notes

Loading notes...