Mind the Blind Spots: A Focus-Level Evaluation Framework for LLM Reviews

TL;DR

This paper introduces a focus-level evaluation framework to systematically assess whether LLM-generated reviews attend to the same critical facets as human experts, revealing biases in focus areas like technical validity and novelty.

Contribution

It proposes a novel focus-level evaluation framework and pipeline for assessing LLM reviews across predefined facets, addressing a gap in systematic review trustworthiness evaluation.

Findings

LLMs focus more on technical validity than novelty.

LLMs tend to overlook aspects like novelty in reviews.

Focus distributions differ significantly between LLMs and humans.

Abstract

Peer review underpins scientific progress, but it is increasingly strained by reviewer shortages and growing workloads. Large Language Models (LLMs) can automatically draft reviews now, but determining whether LLM-generated reviews are trustworthy requires systematic evaluation. Researchers have evaluated LLM reviews at either surface-level (e.g., BLEU and ROUGE) or content-level (e.g., specificity and factual accuracy). Yet it remains uncertain whether LLM-generated reviews attend to the same critical facets that human experts weigh -- the strengths and weaknesses that ultimately drive an accept-or-reject decision. We introduce a focus-level evaluation framework that operationalizes the focus as a normalized distribution of attention across predefined facets in paper reviews. Based on the framework, we developed an automatic focus-level evaluation pipeline based on two sets of facets: target (e.g., problem, method, and experiment) and aspect (e.g., validity, clarity, and novelty), leveraging 676 paper reviews (https://figshare.com/s/d5adf26c802527dd0f62) from OpenReview that consists of 3,657 strengths and weaknesses identified from human experts. The comparison of focus distributions between LLMs and human experts showed that the off-the-shelf LLMs consistently have a more biased focus towards examining technical validity while significantly overlooking novelty assessment when criticizing papers.