This is a very good post: New blog post: the state of AI safety in four fake graphs.

→ View original post on X — @sama,