@repligate @AdriGarriga @MInusGix I am in agreement with you that certain types of optimization pressure toward certain types of alignment targets are indeed probably net-negative…
Posters: davidad. davidad
2 likes recorded by the source corpus.
Topics: AI safety