@Algon_33 @ESYudkowsky 1. Value is fragile and hard to specify2. Corrigibility is anti-natural3. Pivotal processes require dangerous capabilities4. Goals misgeneralize out of distribution5.…

Posters: davidad. davidad

3 likes recorded by the source corpus.

Topics: AI safety

View Post: https://x.com/davidad/status/1660547794627641345