TY  - JOUR
AU  - Luo, Xiaoliang
AU  - Rechardt, Akilles
AU  - Sun, Guangzhi
AU  - Nejad, Kevin K.
AU  - Yáñez, Felipe
AU  - Yilmaz, Bati
AU  - Lee, Kangjoo
AU  - Cohen, Alexandra O.
AU  - Borghesani, Valentina
AU  - Pashkov, Anton
AU  - Marinazzo, Daniele
AU  - Nicholas, Jonathan
AU  - Salatiello, Alessandro
AU  - Sucholutsky, Ilia
AU  - Minervini, Pasquale
AU  - Razavi, Sepehr
AU  - Rocca, Roberta
AU  - Yusifov, Elkhan
AU  - Okalova, Tereza
AU  - Gu, Nianlong
AU  - Ferianc, Martin
AU  - Khona, Mikail
AU  - Patil, Kaustubh R.
AU  - Lee, Pui-Shee
AU  - Mata, Rui
AU  - Myers, Nicholas E.
AU  - Bizley, Jennifer K.
AU  - Musslick, Sebastian
AU  - Bilgin, Isil Poyraz
AU  - Niso, Guiomar
AU  - Ales, Justin M.
AU  - Gaebler, Michael
AU  - Ratan Murty, N. Apurva
AU  - Loued-Khenissi, Leyla
AU  - Behler, Anna
AU  - Hall, Chloe M.
AU  - Dafflon, Jessica
AU  - Bao, Sherry Dongqi
AU  - Love, Bradley C.
TI  - Large language models surpass human experts in predicting neuroscience results
JO  - Nature human behaviour
VL  - 9
IS  - 2
SN  - 2397-3374
CY  - London
PB  - Nature Research
M1  - FZJ-2025-02220
SP  - 305 - 315
PY  - 2025
AB  - cientific discoveries often hinge on synthesizing decades of research, a task that potentially outstrips human information processing capacities. Large language models (LLMs) offer a solution. LLMs trained on the vast scientific literature could potentially integrate noisy yet interrelated findings to forecast novel results better than human experts. Here, to evaluate this possibility, we created BrainBench, a forward-looking benchmark for predicting neuroscience results. We find that LLMs surpass experts in predicting experimental outcomes. BrainGPT, an LLM we tuned on the neuroscience literature, performed better yet. Like human experts, when LLMs indicated high confidence in their predictions, their responses were more likely to be correct, which presages a future where LLMs assist humans in making discoveries. Our approach is not neuroscience specific and is transferable to other knowledge-intensive endeavours.
LB  - PUB:(DE-HGF)16
C6  - 39604572
UR  - <Go to ISI:>//WOS:001365146700001
DO  - DOI:10.1038/s41562-024-02046-9
UR  - https://juser.fz-juelich.de/record/1041321
ER  -