It's all too easy to derail AI chats.

Fun Fact: Since the llm itself does not remember anything the chat system sends the entire chat to the ai to process which when there is a lot of text can make it derail very fast

I want to let it derail