New paper: LLMs falter at probabilistic belief updating, not action optimization
Dr_Atoosa · x · 2026-10-23
A new paper from Jessica Hullman's group examines why LLMs make bad calls as decision assistants (risk estimation, treatment recommendation): did the model fail to arrive at correct beliefs, or fail to optimize actions under the right utility? Findings and independent replication by xuanalogue suggest LLMs are worse at probabilistic belief updating than at optimizing under correct, usually deterministic, beliefs.
More from Research
- Google says Gemini 4 Argon found fusion counterexamples and designed 5 lab-verified enzymes in hours — Eric520CC · 2026-10-24
- MIT left 13 AI agents alone on virtual islands for 30 hours — they started to play, and it made them better — alliekmiller · 2026-10-24
- Robert Plenge: How much causal human biology we know determines AI's role in drug discovery — rbhar90 · 2026-10-24
- Bug Hunt Bench: blind-graded leaderboard pits frontier coding models against 105 real planted bugs — PawelHuryn · 2026-10-24
- How visual evidence gets encoded: different priors interpolate into different hallucinations — debreuil · 2026-10-24
- MIT Team: Only 2 of 286 TB Antigen Candidates Detected on MHC-I — BrianHie · 2026-10-24