Ever seen how GPT sections its responses when explaining something (ie. studying maths or biology). Many times you want to go deeper into one of those sections, without moving the conversation any further. Like a post-it note, or a sub-section. You could explore deeper into what GPT explained, effectively turning that section into its own inner-chat, and rendering the conversation into a directory tree-like structure (now that I mention it, that would be sick to visualize it like that once implemented).
This would de-clutter user’s dashboard and improve the quality of the conversations. Personally, it would save me a lot of time spent on topic exploration and conversation organization.
PD. If you like this, please do the visualization tree thiny, it would be pretty cool
Thank you Eric!
Yes. /btw and /fork are known commands. But many, many people don’t use Codex and never go beyond the context of GPT chats. Right now, to me at least, my chat list feels very untidy. If you think about it GPT is missing that “forking” capacity outside of Codex. Not everyone writes code/scripts, not everyone is a developer. Many students use this to go deep into topics while never touching google, SO, quora, etc and maintaining everything on the same place. I believe the capacity to go laterally on chats would be good addition.
I didn’t know about that. I tried it and it misses a few points. It copies the entire original chat into the new one, it is added to the side bar as a separate unit (except for its label), it is a branch of the entire chat up to the last response.
My request is the “branch” to start from a section of a response. In this sense, the word “fork” ends up being illustrative, because I don’t need the entire chat again, only the relevant section.
Example
Here I am learning about quadratic functions. You can see there is 2 separate sections in the screenshot. I might want to ask follow ups to the lower section (“Caso A”), and only require that for context. Not needing the WHOLE conversation again (like a branch). See, I don’t want or need infinite branches for my follow ups (cluttering my side bar). I am just making precise questions to clarify the received information. UI-wise this could have a lateral flow so I could review all my follow ups quickly inside the conversation’s context. It would feel like a notebook with annotations.
I believe branches are a separate use case. 1) I might be exploring ideas and have 2+ possible directions → I branch. Chat direction is FINE, I just need specific follow ups without losing the main thread ( I prefer the received information to be organized) → I ask a folllow-up from the relevant section and I can go back to main thread when done
ERRATA:
I believe branches are a separate use case. 1) I might be exploring ideas and have 2+ possible directions → I branch. 2) Chat direction is FINE, I just need specific follow ups without losing the main thread ( I prefer the received information to be organized) → I ask a folllow-up from the relevant section and I can go back to main thread when done
