← Back
Empirical Priors
Writing
Long-form thinking on medicine, machine learning, and the space between them.
May 2, 2026/Differential, Issue 01
The Word They Cut→
Close read of Brodeur et al. (Science 2026), the latest "AI matches doctors" headline. The 78/80 R-IDEA score reads more as a contamination signal than a capability claim. Eight methodological problems below.
Clinical AILLMMethodologyDifferential Series
empiricalpriors.io