Half of peer reviewers now use AI: survey

Half of scientific peer reviewers now rely on artificial intelligence to evaluate manuscripts, yet 75% of researchers report they cannot tell if journals use the technology themselves. A survey of 1,600 academics recently conducted by the open-access publisher Frontiers found that 53% of reviewers have incorporated AI into their workflow, with 24% increasing use over the past year. Among early-career researchers, 87% reported using AI for manuscript evaluation.

AI adoptions across tasks, regions: Reviewers say they use AI primarily to draft comments, summarize manuscripts and check language clarity. Some also apply AI to technical evaluation, including reviewing statistical analyses, methods, and experimental design. Adoption varies by region, with 77% of reviewers in China reporting AI use compared with lower rates in North America and Europe. Reviewers in disciplines with high submission volumes or complex research cite AI as a tool to manage repetitive tasks and maintain consistency across manuscripts.

Limited awareness of publisher AI practices: Despite widespread reviewer adoption, most researchers reported they cannot tell whether journals employ AI in editorial or production workflows. Policies and practices differ across publishers, disciplines and regions. Some journals have introduced in-house AI platforms or require disclosure when external tools are used, but adoption is inconsistent. Researchers often have no information on whether AI influences manuscript handling or editorial decisions.

Reviewers’ operational contrasts: Reviewers report faster evaluations, clearer feedback and reduced repetitive work when using AI. The survey shows that AI use in peer review is largely invisible to the broader research community, revealing a gap between reviewer practices and publisher transparency. Kamila Markram, CEO and co-founder of Frontiers, noted that “while AI is steadily finding its place in peer review, its full potential is currently untapped” as it is often used for “surface tasks, like polishing language, drafting text, or handling administration, rather than for deeper analytical and methodological work.”