Alignment pretraining: AI discourse creates self-fulfilling (mis)alignment

Comments

from Hacker News https://ift.tt/lcoZKAQ

Comments