causal
Reasoning creates flywheel for frontier labs
Reasoning models create a data flywheel for frontier labs because user answers that people consistently like can serve as verifiable rewards and be fed back into the model, unlike pre-reasoning models where RLHF from social-media vibes was hard to use.
causalpending
Speaker
Gavin BakerEvidence Quote
“if a lot of people are asking a similar question, they're consistently either liking or not liking the answer, then you can use that as a verifiable reward”
Created: 11/20/2026, 2:50:27 AM
Your notes
Loading notes...