Summary of ‘quis Custodiet Ipsos Custodes?’ Who Will Watch the Watchmen? on Detecting Ai-generated Peer-reviews, by Sandeep Kumar et al.
‘Quis custodiet ipsos custodes?’ Who will watch the watchmen? On Detecting AI-generated peer-reviews
by Sandeep Kumar, Mohit Sahu, Vardhan Gacche, Tirthankar Ghosal, Asif Ekbal
First submitted to arxiv on: 13 Oct 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Digital Libraries (cs.DL); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper introduces two novel approaches to detect whether a peer-review was written by ChatGPT or not: the Term Frequency (TF) model and the Review Regeneration (RR) model. The TF model is based on the idea that AI-generated text often repeats tokens, while the RR model leverages the fact that ChatGPT generates similar outputs when re-prompted. Both models are stress-tested against token attacks and paraphrasing, demonstrating robust performance in detecting AI-generated reviews. Furthermore, the paper proposes a defensive strategy to mitigate the impact of paraphrasing on these detectors. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper aims to solve a real-world problem by helping editors or chairs determine whether peer-reviews were written by ChatGPT or not. It introduces two new methods: Term Frequency (TF) and Review Regeneration (RR). The TF model looks for repeated tokens in AI-generated text, while the RR model checks if reviews are similar when re-prompted. These models are tested against fake attacks and paraphrasing, showing they can accurately detect AI-written reviews. |
Keywords
» Artificial intelligence » Token
