Mathematician Michael Goldman: trusts LLM reviews over human peer review

skdh · x · 2026-11-18

Mathematician Michael Goldman published a provocative piece titled "Do We Still Need Math Journals?" — his answer is no. He says he now trusts LLM-based reviewing more than human peer review, including his own.

He also ran the numbers: French public research spends about €78M per year on journal subscriptions, half of it going to Elsevier — enough to give every one of France's 100-200k researchers a basic LLM subscription with €20M left over.

With dissemination covered by arXiv, quality control by AI, and selection by informal peer discussion (while formal publication takes over a year), journals' only remaining function is counting papers for hiring and promotion — a €78M/year system whose final output is a KPI.

Original post →

More from AGI Musings

AGI Musings channel →