This is a wonderful Xmas present for me: the most articulate description of an alignment-oriented approach for raising models that might work. The only thing I would add is that alignment is…
Posters: eshear. eshear
2 likes recorded by the source corpus.
Topics: AI safety