AI13 Rethinking peer review in the artificial intelligence era: artificial intelligence adoption for dermatology peer review

Abstract

Inspired by discussions within the BAD editorial office, we conducted a multiplatform poll to capture stakeholder views on the feasibility, acceptability and perceived limitations of artificial intelligence (AI) in peer review. This study examines attitudes towards the use of AI in scholarly peer review, motivated by growing interest in the role of AI across dermatology and research publication. Polls were posted between September and October 2025 across the social media channels of three BAD-affiliated journals – Clinical and Experimental Dermatology, British Journal of Dermatology and Skin Health and Disease – with a combined follower base of approximately 28 684. Participants were asked about their general stance on AI in peer review, its potential utility, current use, whether AI-only evaluation constitutes peer review, and how AI may influence future research and publishing workflows. Nearly half of respondents (47.1%) expressed support for AI in peer review, with 76.6% agreeing that AI could be used within the process. Despite this optimism, uptake remains low, with only 27.5% reporting active use of AI. Notably, only 15.7% felt that AI alone constitutes peer review, underscoring a broad consensus that human judgement remains essential. Qualitative thematic analysis revealed cautious optimism. Respondents viewed AI as a potential tool for improving efficiency – particularly in detecting plagiarism, checking formatting and supporting administrative tasks – while consistently emphasizing concerns regarding confidentiality, bias, transparency and contextual interpretation of academic work. These findings mirror wider scepticism within scientific publishing, where leading journals have restricted AI in peer review due to ethical and legal risks. AI has the potential to enhance the peer review and publication process by streamlining repetitive tasks such as checking word limits, verifying figures and ensuring referencing accuracy. Core evaluative judgements – novelty, methodological soundness and significance – should be reserved for human experts.