Concern alignment is a new diagnostic framework that audits AI peer reviews at the level of individual concerns via match graphs to reveal gaps in detection, calibration, and prioritization beyond verdict agreement.
ReviewEval: An evaluation framework for AI-generated reviews
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
years
2026 3verdicts
UNVERDICTED 3representative citing papers
PRAIB reveals LLM reviews are less variable, positively biased, overconfident, longer, and overlook atomic weaknesses noted by humans compared to real reviewer feedback.
PRISM benchmark finds LLMs match or exceed humans on isolated review dimensions like novelty verification but none achieve the balanced performance of human reviewers across depth, flaw prioritization, and constructiveness.
citing papers explorer
-
What Makes a Good AI Review? Concern-Level Diagnostics for AI Peer Review
Concern alignment is a new diagnostic framework that audits AI peer reviews at the level of individual concerns via match graphs to reveal gaps in detection, calibration, and prioritization beyond verdict agreement.
-
PRAIB: Peer Review AI Benchmark of Behaviour of LLM-Assisted Reviewing
PRAIB reveals LLM reviews are less variable, positively biased, overconfident, longer, and overlook atomic weaknesses noted by humans compared to real reviewer feedback.