NEJM AI editorial argues trial design is key to validating model-guided interventions
zakkohane · x · 2026-11-19
A new NEJM AI editorial by Joseph W. Hogan argues trial design is fundamental to understanding not just whether a model-guided intervention works, but how. ML researcher zakkohane shared it with a self-deprecating jab that the statistician gives "grubby ML researchers" more credit and less tough love than they deserve — a window into ongoing tension between medical statistics and ML evaluation culture.
More from AGI Musings
- Engineer's rule for the AI era: do key tasks manually first, then offload to agents — jmschreiber91 · 2026-11-19
- Washington Post op-ed calls for a national AI pause, says China-race fear is overblown — wfithian · 2026-11-19
- Engineer: a decade of low-level skills is now delegable to agents — grieve it — jmschreiber91 · 2026-11-19
- PhD pipeline criticized: academia 'needs its cheap labour' as graduates find no one hires their expertise — miniapeur · 2026-11-19
- Ex-Anthropic researcher Jacob Coxon to protest outside former workplace on IPO day, calls for ban on 'AIs building better AIs' — Polymarket · 2026-11-19
- Philosophers argue LLMs miss the point on consciousness: the common denominator is life, not language — erikphoel · 2026-11-19