The world of academic publishing is undergoing a quiet revolution, one that threatens to undermine the very foundation of scholarly integrity. As AI continues to permeate every aspect of our lives, it's no surprise that it has found its way into the heart of academic journals. The recent case of a retracted article in the Journal of Medical Ethics highlights the growing concern over the use of AI-generated content and the challenges it poses to the detection systems of academic journals.
The article in question, "Panem, Corticoids and Circenses: the ethical fallout of Enhanced Games," by Alexis Demas, was found to contain multiple citations to "bibliographic fictions," which are seemingly the result of unacknowledged AI use. The author, who used a Yahoo email address and whose institutional affiliation could not be verified, had submitted a manuscript that was eventually retracted. This incident sparked a debate among scholars and editors about the ethical implications of AI in academic writing and the need for improved detection systems.
Udo Schuklenk, a professor of philosophy at Queen's University, described the situation from the editor's perspective. He noted that the number of made-up references in the Journal of Medical Ethics article suggests that the author was careless enough to leave a plethora of fake references in the manuscript. Schuklenk emphasized that this doesn't happen by accident and that during the proof corrections stage, the author had ample time to correct any errors but chose not to. He also mentioned the lack of detection tools available to assist in making reliable determinations of AI use.
The core issue, according to Ognjen Arandjelović, is not about helping authors avoid mistakes but making it costly to submit work with citations that do not exist or do not support what they are cited for. Arandjelović argues that the challenge lies in deterrence and accountability rather than assistance and coaching. He criticizes the tendency to treat reference failures as a mechanistic problem that can be addressed through technology and bureaucracy, suggesting that this approach misdiagnoses the failure mode.
Schuklenk's editorial highlights the importance of sophisticated automated reference checks in detecting AI-generated content. He praises Wiley, the publisher of Bioethics, for implementing such a system, which would have eliminated the manuscript after screening for fake references and AI-generated content. In contrast, he criticizes the BMJ group of journals for not possessing similar capabilities or not deploying them effectively.
The incident raises deeper questions about the role of reviewers and editors in maintaining academic integrity. Schuklenk acknowledges the difficulty of finding willing reviewers for the ever-increasing number of papers, especially given that they are expected to work pro bono. He suggests that the lack of detailed reference checks by reviewers is understandable, given that publishers could have automated systems in place to undertake such tasks.
In conclusion, the case of the retracted Journal of Medical Ethics article underscores the urgent need for improved detection systems and a reevaluation of the role of technology and human oversight in academic publishing. As AI continues to evolve, so must the mechanisms for ensuring the integrity and authenticity of scholarly work.
