The peer review process is a unique testing ground for AI integration. AI can assist peer review by triaging submissions, detecting manipulations, and supporting reviewers with language and structure, provided maintenance of robust human oversight. However, it presents serious risks from fueling paper mills and plagiarism to amplifying bias, generating hallucinated references, breaching confidentiality, and fostering over-reliance that weakens critical judgment. Despite these challenges, it is important to embrace its benefits responsibly while safeguarding research integrity to rethink the future of peer review.
The interactive decision tree below provides a roadmap for journals to navigate the readiness in adopting AI through various aspects including clear policies, transparency, safeguards, infrastructure, and training.
