4 ms·
I think the assumption is we can align models less intelligent than ourselves, the hard part is aligning models that are more.
by Ninjinka 3y ago
I think the assumption is we can align models less intelligent than ourselves, the hard part is aligning models that are more.