It’s going to be significantly faster very soon, we have seen how AlphaGo evolve... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		renonce on July 25, 2024 \| parent \| context \| favorite \| on: AI solves International Math Olympiad problems at ... It’s going to be significantly faster very soon, we have seen how AlphaGo evolved into KataGo which is many magnitudes more compute efficient

thomasahle on July 26, 2024 [–]

The main difficulty to scaling Alpha Proof is finding theorems to train it with. AlphaGo didn't have that problem because it could generate it's own data.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact