its actually a fairly good test case. "dont make an AI that doesnt use the n-word" is clear value alignment represented by a single, unambiguous rule and as usual an entire industry of Value Experts…

Posters: eigenrobot. eigenrobot

7 likes recorded by the source corpus.