[zukzuk]

I find that Claude almost always ends its response with some sort of follow up question, despite my system prompt telling it not to.

I never really used ChatGPT much though so maybe Claude is just relatively less egregious?