Northwestern Prof: Even Small Open-Source Models Write More Informative Peer Reviews Than Humans

JessicaHullman · x · 2026-11-16

Jessica Hullman (Northwestern) doesn't dispute that AI can write more informative peer reviews — her lab measures reviews' predictiveness of downstream signals like citations and finds even small open-source models beat humans on average. But she argues handing review fully to AI removes human values from science, and questions whether that's sustainable.

Related event: Professor: AI Reviews May Be More Informative but Risk Science's Human Value(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →