@peterwildeford IMO it's a given that some future AIs will be deceptive, so it's more important to understand what deceptiveness looks like in an AI than to try to prevent it happening.

Posters: timfduffy. timfduffy

2 likes recorded by the source corpus.