While I don’t think that this alone is enough, I do think it’s a good addition to many other plans. Having a time-bounded run is compatible with many other ideas, like optimizing for Obedience to the Operator or corrigibility, etc.
While I don’t think that this alone is enough, I do think it’s a good addition to many other plans. Having a time-bounded run is compatible with many other ideas, like optimizing for Obedience to the Operator or corrigibility, etc.