Study Finds LLMs Infer Drug Class from Suffixes, Not Knowledge
allen_ai · x · 2026-08-19
Researchers at UT Austin and partners found that LLMs often lack drug-specific knowledge and instead rely on the morphology of drug names—such as suffixes like -pril or -olol—to infer answers. In tests with Olmo 3, 51–59% of drugs showed little sign of specific knowledge, while another 12–18% appeared affix-driven. Because Olmo 3 is fully open, the team traced this behavior back to the training data, revealing the source of this shortcut and its implications for health-related queries.
Related event: LLMs fake drug knowledge by guessing from name suffixes, study finds(2 posts)→
More from Models
- Prediction: frontier AI revenue growth could slow even as AI usage keeps rising — yi111 · 2026-10-16
- Meta teased 816B-param open-weight Muse Spark Open with 1M context, says meme thread — alexandr_wang · 2026-10-16
- Nolla Health Post-Trains Nemotron 3.5 VL: Visual Diagnosis Jumps from 35% to 66.8% — LoganGrasby · 2026-10-16
- Dev argues coding models now beat nearly all engineers in every branch of software — zack_overflow · 2026-10-16
- Prediction: Chinese frontier models could match GPT-6 by March 2027 — teortaxesTex · 2026-10-16
- Liquid AI's d1-ultra, #1 on Artificial Analysis Decision Index, lands on Vercel — JosephJacks_ · 2026-10-16