Philosophy professor envisions per-reviewer AGENTS.md: AI agents extending the will of peer reviewers

JacksonKernion · x · 2026-09-21

Philosophy professor Jackson Kernion sketches LLM-assisted peer review: reviewers can and should be opinionated. He imagines each maintaining their own AGENTS.md, skills and files, with automated review as a starting point — per-reviewer defaults could even be generated from past review exchanges.

The premise: if you can articulate what you look for in articles, with examples and reasons, agents can extend the reviewer's will. He rejects a particularist view that each philosophical work must be evaluated on its own terms. He concedes the practicalities are very hard, possibly impossible — more likely individuals become "super-reviewers," or publishing stays like art curation. Context: a journal editor told him desk-rejecting bad papers is easy; finding good reviewers is the hard part.

Related event: Philosopher Proposes Personalized Review Agents via AGENTS.md(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →