'Quis custodiet ipsos custodes?' Who will watch the watchmen? On Detecting AI-generated peer-reviews

TL;DR

This paper introduces two models, TF and RR, to detect AI-generated peer reviews, demonstrating their effectiveness and robustness against attacks, thereby aiding editors in maintaining review integrity.

Contribution

The paper presents novel detection models specifically designed for identifying AI-generated peer reviews, addressing real-world challenges and improving over existing generic detectors.

Findings

RR model is more robust against paraphrasing attacks

Both models outperform existing AI text detectors

TF model performs better without attacks

Abstract

The integrity of the peer-review process is vital for maintaining scientific rigor and trust within the academic community. With the steady increase in the usage of large language models (LLMs) like ChatGPT in academic writing, there is a growing concern that AI-generated texts could compromise scientific publishing, including peer-reviews. Previous works have focused on generic AI-generated text detection or have presented an approach for estimating the fraction of peer-reviews that can be AI-generated. Our focus here is to solve a real-world problem by assisting the editor or chair in determining whether a review is written by ChatGPT or not. To address this, we introduce the Term Frequency (TF) model, which posits that AI often repeats tokens, and the Review Regeneration (RR) model, which is based on the idea that ChatGPT generates similar outputs upon re-prompting. We stress test these detectors against token attack and paraphrasing. Finally, we propose an effective defensive strategy to reduce the effect of paraphrasing on our models. Our findings suggest both our proposed methods perform better than the other AI text detectors. Our RR model is more robust, although our TF model performs better than the RR model without any attacks. We make our code, dataset, and model public.