Reaction
Retrieve-then-verify medical fact-checking reportedly fails in systematic ways
A post says bigger models, more reasoning, better sources and medical fine-tuning don’t fix these failures.
TLDR
An October 8 post announces a NeurIPS GenAI4Health oral presentation for research led by Heyuan Huang, placing it in the top 4% of roughly 250 submissions. The author says retrieve-then-verify medical fact-checking fails in systematic ways that bigger models, more reasoning, better sources and medical fine-tuning don’t fix.
Combined views
70
1 Source, first seen ago