I just want to verify that I understand how the assistant role works, if I were, for example, doing a ChatGPT style implementation:
I set the System role if I want, and then my prompt goes in the “user” role message. Then ChatGPT responds. If I continue the conversation, I send back the system role message, my previous user message, and then populate the assistant role message with its previous response, then add my new prompt as a user role message after that, and this continues back and forth.
Is that, more or less, correct?
sounds good 
but you can send back whatever and whenever you like or even change the system message whenever you like. Just go into the playground and try yourself.
Yeah, I knew I could change the System message. I just wanted to make sure I understood the basic idea properly. My plan is to make the chat history editable.
Thanks