I find it very interesting that “aligning with human desires” somehow includes prevention of a human trying to bypass the safeguards to generate “objectionable” content (whatever that is). I think the “safeguards” are a bigger problem with aligning with my desires.
replies(4):