Artificial intelligence and scientific peer review: usage models, policies, and challenges

Abstract

The aim of this article is to critically analyze the latest trends in the use of artificial intelligence (AI) within the scientific peer review process. Both the benefits and risks associated with the application of AI tools by reviewers are discussed. First, AI is presented as a linguistic aid—for example, in stylistic proofreading and
rewording review content. Next, AI is analyzed as substantive support in evaluating manuscripts, including the capabilities and limitations of AI systems in formulating conclusions and detecting scientific problems. Subsequently, ethical issues are explored, such as the reviewer's responsibility for AI-assisted content,
the need for transparency regarding the use of these tools, the risk of biases encoded in the models, and potential abuses (e.g. generating fake citations). The following segment outlines the current guidelines of leading publishers and institutions (including Elsevier, Springer Nature, COPE, Nature, and ICMJE)
regarding the use of AI by reviewers, with a particular focus on confidentiality principles and the obligation to disclose AI usage. Finally, the problem of epistemic inequalities is addressed – specifically linguistic and infrastructural barriers – affecting the access of reviewers from different regions to advanced AI tools.