The authors argue that generative AI introduces a new class of alignment risks because interaction itself becomes a mechanism of influence. Humans adapt their behavior in response to AI outputs, ...
Results that may be inaccessible to you are currently showing.
Hide inaccessible results
Feedback