4 sources healthy · latest episode 9 Nov 2026
descriptive claim

The speaker says current models perform very well on tasks that can be trained efficiently with reinforcement learning.

Unresolved speaker · Why AI Compute Comes in Rectangles — Erik Bernhardsson, Modal CEO · 22 Oct 2026

In their words

The transcript segments this claim was taken from

“Like models are today are like very good at like doing things that can be trained very efficiently, you know, using reinforcement learning.”
1:23:31Report a misattribution
0.90how strongly it was put, not whether it is true
0.80importance to the conversation
0.95confidence the claim was read correctly