)
)
AI for detecting paper mill papers
An important step towards improving research integrity
Paper mills are a systemic threat to research integrity, contaminating the evidence, influencing citations, and potentially affecting clinical decision making. Paper mills are companies that generate manuscripts engineered to resemble legitimate scientific articles, frequently relying on fabricated, manipulated, or duplicated data and images. Authorship of those manuscripts is sold to authors and submitted to scientific journals for publication as if they were genuine scholarly work. Major media outlets and leading scientific journals increasingly report on the scale of the problem and its implications for trust in science. Yet detection of papers originating from paper mills remains challenging for several reasons. Firstly, these papers are dispersed across journals and are typically scrutinised individually; secondly, post-publication review (ie, the evaluation of published research by readers and independent experts outside the formal peer review process) relies on a relatively small number of specialised sleuths; thirdly, some journals may lack incentives to acknowledge large scale contamination, and in some instances editorial involvement has been reported; and finally, many researchers remain unaware of this problem.