A Missing Piece for Trustworthy AI Reviewers: From Benchmarking Rhetorical Robustness to SciCore Review

What happened: Researchers introduce SciCore, a dual-branch AI reviewer that averages a full-manuscript judgment with a judgment based on an extracted, structured science core, aimed at reducing sensitivity to rhetorical variation in manuscripts. Key numbers: RobustReview Why it matters: The benchmark reveals false robustness, where low rewrite sensitivity coincides with score collapse across papers, and shows human alignment and rhetorical robustness rank reviewers differently. SciCore achieves a leading joint stability-discrimination profile while maintaining competitive human alignment.

Researchers introduce SciCore, a dual-branch AI reviewer that averages a full-manuscript judgment with a judgment based on an extracted, structured science core, aimed at reducing sensitivity to rhetorical variation in manuscripts.

The benchmark reveals false robustness, where low rewrite sensitivity coincides with score collapse across papers, and shows human alignment and rhetorical robustness rank reviewers differently. SciCore achieves a leading joint stability-discrimination profile while maintaining competitive human alignment.