I’m not sure whether to call this a feature request or a bug.
Deep Research provides many markers within the text response to use as pointers/citations for where it found information. Here is an example: 【45†L75-L83】.
ChatGPT says: The marker “【45†L75-L83】” is a citation reference. It indicates that the information just mentioned is supported by content found in source number 45, specifically lines 75 through 83 of that source. This citation style helps trace the origin of the content for verification or further reading.
Here is the problem. You can see while chatGPT does its Deep Research, it shows the progress of the sources it is referencing in a separate side window. The information in that window does not persist, meaning the citation markers are now just junk in the responses.
Three problems arise from this.
- Now you have responses with junk in them.
- We all really need a way to use these markers to verify accuracy of responses and for whatever other purposes.
- Then there is the question of do you want a version of the response that doesn’t have the markers.
How can the citations be recovered?
How can we keep citations where they persist until chats are permanently deleted?
How can citations be removed, toggled or some method to utilize them better.
It would be great to have them embedded in a response as an appendix, in a similar way that it would be useful to have an outline for a response.