Comment by hhh

this is just what I would expect from a solid prompt for an LLM to act a certain way? I was using gpt-3 around its release to get similar kinds of behavior for chatbots, did we lose another one to delusion?