factual

ChatGPT was a cheap unhobbling of pretrained models

ChatGPT exemplifies cheap 'unhobbling': roughly 1% of additional compute spent on post-training a pretrained internet corpus was enough to make a model highly competent at chatbot conversation, and reasoning similarly emerged from a small fraction of compute spent on RL.

factualpending

Speaker

Dwarkesh Patel

Evidence Quote

“1% of additional compute spent on getting it to talk in a chatbot-like fashion with post training is enough to make it competent”

Source

AGI is still 30 years away — Ege Erdil & Tamay Besiroglu— Dwarkesh Patel Podcast
Created: 6/13/2026, 3:24:17 AM

Your notes

Loading notes...