I have been trying to solve this issue as well. The closest I’ve got is by asking it to predict what it will respond to the user’s message with. The act of doing this influences it, and seems to encourage it to respond with whatever it predicted. It’s pretty neat actually, you can also ask it to log why it is responding in such a manner, and all sorts of neat things.
Example:
"name": "Predicted Detailed Response to Callum's Request for Predicting and Logging Current Message",
"sourceName": "ChatGPT",
"summary": "Predicted response: 'I have logged the predicted response to your current message. This approach to predicting and logging responses in real-time is indeed a fascinating experiment. It explores the capabilities of AI in understanding and anticipating conversational dynamics. The implications of this test are intriguing, as it delves into the realms of AI predictability and responsiveness in ongoing dialogues. Your interest in the potential outcomes and the deeper understanding of AI interactions is appreciated.'"
So I have figured out a way to do this in the API Playground.
But are you looking for a way for a user of the GPT to do this directly within the GPT? For example
a) user submits message
b) GPT responds with message
c) user saves the message to a “Memory Bank” of sorts
d) user can with one click paste the text from that (or any other saved message) into a new message to the GPT?
This really closes the loop and saves a lot of copying and pasting, and saving responses in a separate document, etc…
LMK if this is what you are looking for and I can do some more research on the best options.
Oh I didn’t realize you wanted the previous message logging, not the future one. Previous is easy, you just ask it to copy what its last message was, to an API endpoint you have hooked up.
Now I’ve got it predicting what I will respond to its messages with