I think we would have better results and the need for less computing if before generating the text we could have a phase to actually select the relevant articles ourselves. Like submit question, get papers, select papers then let the llm generate an answer based on that. It feels very wrong to let the machine select the papers, especially without a view on what it is doing and how it is doing its 8 paper cherry picking.
I agree - what criteria does elicit use to select the top 8 papers? Would be great to be able to upload paper first and use that to explore citation networks - define relevant question
![Things you don't say out loud in academia [9 open secrets]](http://i.ytimg.com/vi/sgAlFPMugWw/mqdefault.jpg)
Brilliant. Would've saved me three years off my PhD, so that's... not totally frustrating at all. But... brilliant.
Does your database also include scientific posters and abstracts?
Unfortunately not. Do you know if these are easily accessible or published in one place?
Where is the question brainstorm feature? it was a great, but I don’t see in the new version.
We're in the process of reintegrating it!
Hi, why my export is not organized well in excel? The columns that I added do not appear as columns
There was a bug with this but the fix has been implemented and should be out shortly!
This is fixed now!
I think we would have better results and the need for less computing if before generating the text we could have a phase to actually select the relevant articles ourselves. Like submit question, get papers, select papers then let the llm generate an answer based on that. It feels very wrong to let the machine select the papers, especially without a view on what it is doing and how it is doing its 8 paper cherry picking.
Is this about the summary at the top? If so, this is on deck.
I agree - what criteria does elicit use to select the top 8 papers? Would be great to be able to upload paper first and use that to explore citation networks - define relevant question
Fascinating 🎉