Getting the source of a response


I am trying to write a prompt that curates medical literature and need the response to provide the sources. Ideally, I’d like to get a one-to-one relationship in that for every response text I also get it’s source document. For example,

Find references
What is the efficacy of oral Fisetin in pruning senescent cells as a Senolytic?


  1. Kim, Y.H. et al. Fisetin is a senotherapeutic that extends healthspan, slows aging, and increases resistance to oxidative stress and chemotherapeutics. Cell Metab. 24, 675–691 (2016).

I’m getting quality response text but need the source document to read more and/or use it to create additional prompts.

Any assistance would be appreciated.


1 Like

@m-a.schenk nailed it. I think the only surefire way you could do this is by using the metadata field in an uploaded file.

1 Like

I dont think GPT3 is gonna help you with that, especially the completion engine. You might wanna be looking in the recently released biomedical explorer by google that can answer this sort of questions.

1 Like