Has anyone used elicit.com to factorize the causal chains for controversial fields/claims? (and can it extrapolate?) Is it worth paying?

[Alex K Chen]

I didn't find the first version very useful, but there are now later versions.. You can create an automatic table of studies showing separate cohorts (like age cohorts) by the columns they give, which is not super-useful.

Eg, if omega 6's are pro or anti longevity? Under separate D6D genes? (prompting).. [they could break down the columns into **different omega 6**'s, rather than the standard demographic variables they're trying to break down...

Or saturated fats from cacao? Or are saturated fats from milk fats *more* healthy because they also have odd-chain SFAs, or less healthy b/c they're animal-based, which mixes them in with pro-mTOR milk proteins? If one is asking saturated fat, they ought to factorize for CONFOUNDERS (like if MUFAs are also being eaten at the same time - Physionic has indicated that MUFAs reduce SFA-induced insulin resistance).

Or vegetarianism on longevity (effect size is more clear, but it's still highly debated) Or *what spices can be predicted to have the highest anti-glycation potential*?

Or the ideal dose of Adderall [including microdoses] to minimize long-term neurotoxicity?

Or pro-aging/anti-aging marks of all the different histone methylations. Or which kinds of damage to DNA are the most repairable and which are the least (think DNA-protein adducts, which is in the literature, but would elicit.org guess it, especially b/c it still has difficulty accessing paywalled articles unless uploaded?)

More samples here: https://www.lesswrong.com/posts/s5jrfbsGLyEexh4GT/elicit-language-models-as-research-assistants?commentId=Cw3WaSKs7pnAvrGYC

MORE:

https://www.science.org/doi/10.1126/scitranslmed.abo2044

Edited January 27 by InquilineKea

[Todd Allen]

To decide if elicit is worth the cost you might try comparing it to free options such as https://consensus.app/search/ and https://www.semanticscholar.org/

Edited September 26, 2023 by Todd Allen

[Alex K Chen]

I just asked "How replicable are papers in autism compred to papers in normal cogsci". It sucked.

But theoretically it could be useful for finding replicability [it can find TO SOME EXTENT contrary results..] But still needs to massively improve

Edited January 27 by InquilineKea