@sendai@eigenmagic.net @futzle@old.mermaid.town Completely agree. With ChatGPT I have a whole bunch of similar directions I've told to commit to memory for responses. It's usually pretty reliable then at stripping it to neutral responses.