The authors selected a set of research papers and used LLMs to rewrite them along two linguistic dimensions derived from the literature on epistemic authority: language proficiency (native vs. non-native English writing) and confidence framing (high vs. low confidence).
LLM peer reviewers reward confident writing, not scientific content
Controlled rewrites of research papers show that large language models systematically prefer high-confidence and native-like linguistic styles, raising concerns about fairness in automated review.
Research Lab
Sai Suresh Macharla Vasu · Sumedh Santosh Joshi · Philipp Christmann · Ivaxi Sheth · Ruta Binkyte · Koel Dutta Chowdhury · +3 more
CISPA Helmholtz Center for Information Security · German Research Center for Artificial Intelligence (DFKI) · University of Technology Nuremberg · INRIA
Research Digest··2 min read
The authors developed a framework to systematically perturb the linguistic style of research papers while preserving scientific content, then evaluated how LLM reviewers scored the resulting versions.
The Daily Edition
The front page in your inbox each weekday morning: the lead, the picks, and why each one earned its place.