AI RESEARCH

Decomposing Physician Disagreement in HealthBench

arXiv CS.AI

ArXi:2602.22758v2 Announce Type: replace We decompose physician disagreement in the HealthBench medical AI evaluation dataset to understand where variance resides and what observable features can explain it. Rubric identity accounts for 15.8% of met/not-met label variance but only 3.6-6.9% of disagreement variance; physician identity accounts for just 2.4