Vulnerable Yet Productive: AI influence in Scientific Publishing

From about three years ago, as a professor of medicine and journal editor, I noticed that there seems to be an increase in timely submissions of written requirements and articles from both medical students and prospective authors. In the past, I saw some honest-to-goodness struggle in writing and expressing scientific content. The challenge to be clear, concise and engaging on top of technical accuracy are real, daunting and will remain so. The fact that more seems to hurdle this challenge can be seen with the rise of AI influence.[1,2]

AI, particularly generative AI, makes it possible to be a bit more efficient to produce manuscripts.[3] Time and effort to generate appropriate sentences in the English language to explain ideas, issues in diagnosis and treatment of diseases, study results, among others, are manageable to meet deadlines without losing much sleep.

While productivity is a huge part of the “publish or perish culture”, the excessive use of AI can render the integrity of scientific publication vulnerable.[4,5] Editors are sharp to see when a manuscript surfaced wholly from the AI belly; luckily this is so far in short supply. Just imagine if more students and authors become slack and just trust AI to churn out these papers for them. The editors will have to work harder to weed out these slackers and protect the integrity of the manuscript generation process.

Editors are aware that more authors have used AI tools to edit their written drafts. There is smoothness in the delivery of sentences which enhances well-researched scientific content they are expressing. AI tools for editing and checking literature are definitely useful in the process. But, as one Japanese medical writer said recently: “Such unintentional AI intervention can thus spread silently and gradually, eroding diversity, rhythm, and personal tone that the original manuscript might have involved.”[6]

How do we help protect the integrity of scientific publications with AI’s growing influence?

Many large and leading scientific journals have recently issued authorship guidelines which includes disclosure of any use of AI tools.[6] While JMUST has not gone to this extent yet (our editors are catching each wanton AI use), we are beefing up our ethical guidelines and continuing our ethics education to protect the integrity of scientific research publication. We also posit that the more diverse the writing style, the more interesting the expression of content.

This present issue features eight clinical and validation studies, one Q-methodology study on lived experiences, four case reports and one viewpoint article. We extend our sincere appreciation to our authors for their contributions, as well as to the JMUST editorial board members and reviewers for their dedicated efforts in evaluating the manuscripts and enabling timely publication of this issue.