Two recent arXiv papers explore the integration of AI into the scientific peer review process, highlighting both its potential benefits and significant risks. One survey synthesizes current techniques for AI-assisted peer review generation, rebuttal assistance, and evaluation methods, aiming to guide practical implementation. The other paper focuses on the security and reliability of AI referees, detailing various attack vectors and experimental findings that reveal vulnerabilities such as susceptibility to prompt injections and biases, questioning the trustworthiness of AI in scientific evaluation. AI
Summary written by gemini-2.5-flash-lite from 3 sources. How we write summaries →
IMPACT AI's role in scientific peer review is under scrutiny, with new research detailing vulnerabilities and potential mitigations for AI referees.
RANK_REASON Two academic papers published on arXiv discuss the use of AI in scientific peer review.