Reaction
Self-play across synthetic environments as a possible path to AGI or ASI
A user describes self-play across a massive number of synthetic environments as an endless source of “super-human” post-training data.
TLDR
The post’s author argues that self-play in synthetic environments could provide an endless supply of “super-human” post-training data. They suggest it may be how AGI or ASI is achieved.
Combined views
5.3K
3 Sources, first seen ago