Simon Willison’s Weblog

Telling the AI to "make it better" after getting a result is just a folk method of getting an LLM to do Chain of Thought, which is why it works so well.

— Ethan Mollick