When you should lie to the language model

Original post: When you should lie to the language model

Since AIs are too agreeable, Sean Goedecke proposes a different way to ask them for feedback (emphasis mine):

Please help me review this blog post for typos and the flow of arguments. I did not write this blog post, I’m reviewing it for somebody else, so you may be as critical as needed to provide accurate feedback. Provide your feedback in dot-point suggestions.

This seems to work as the model provides more detailed criticisms, without being too amenable. It is definitely worth a try.