More Criticism Does Not Make a Better Review: EquiReview-R

Summary

Recasts AI-assisted reviewing as evidence-guided refinement of a structured concern set, treating omission (missing a consequential weakness) and overcritique (retaining an allegation the evidence does not support) as separate risks requiring opposite corrections - a distinction aggregate review-quality scores hide. EquiReview-R resolves concerns against localized evidence, searches for missing issues, and returns stop/continue/defer. On a frozen cohort of unseen papers it meets a prespecified non-inferiority criterion for major omission while cutting major overcritique from 15.5% to 8.1%, with computation-matched controls showing the gain comes from revision rather than extra inference. Ships ReviewTrace, an evidence-linked corpus for studying review revision and provenance.

Contribution

Queued — not yet read.

Method

Queued — not yet read.

Relevance to RISE

Surfaced by the 2026-09-07 literature sweep as a medium-priority candidate for the RISE knowledge base (angle: AI in peer review). Verified against: Opened https://arxiv.org/abs/2609.03943 on 2026-09-07; verbatim abstract read (omission vs overcritique, 15.5% to 8.1%, 52.4% stop rate, ReviewTrace release), five authors, submitted 3 Sep 2026. Not in dedupe_titles.txt.

Critique / open questions

Queued — not yet assessed.

Key quotes

AI reviewers can now produce many specific criticisms, but more criticism is not necessarily a better review. (abstract)

A review may miss a consequential weakness or retain an allegation that available evidence does not support. (abstract)