To the authors

Please clarify the aspects each discussion covers: 1) specialty‑specific vs general medicine, 2) writing vs nonwriting contexts, and 3) current vs future perspectives. Medical AI‑related papers typically target 3 reader types: clinicians well‑versed in AI, those less experienced but interested, and nonclinician AI scientists.² AI‑related publications attract a wider readership than typical medical articles. Browsing these articles, I have noticed many aspects are discussed together, preventing the reader from quickly grasping the paper’s context.

First, clarifying whether AI use is specialty‑specific or broadly applicable is crucial. Studies targeting specific subjects often yield similar results. For instance, humans could not distinguish between an AI‑generated and a human‑written “essay” for “residency applications” in “surgery.”³ Similarly, many studies focus on specific diseases, often presenting similar AI application data, such as diagnostics for diseases A, B, and C. While such targeted data are valuable, authors should clarify whether their findings are confined to specific conditions or generalizable across other areas. For example, discussions regarding the above targets (“surgery” “residency” “essay”) might also apply to manuscript writing more broadly.

Second, differentiating between writing- and nonwriting‑focused AI would aid readers. Tools such as ChatGPT are widely available for writing, while clinical AI adoption requires proven systems. Writing‑focused AI feels more immediate to readers.

Third, the authors should specify whether their focus is on the “current” or the “future.” ChatGPT’s “current” limitations include inaccuracies or contextual misinterpretations, which are likely to improve with technical advancement. Contrarily, AI’s impact on human “future” writing ability and cognition could become a significant problem in the coming years.⁴ Thus, I suggest the authors do the following: 1) clearly indicate which aspects they address, and 2) if their study was targeted, briefly mention broader implications. Stating such a “large frame” of a study may be valuable to readers.

To the journals

Please highlight whether you have journal‑specific regulations on AI use—preferably on the first page of the author guidelines. While many journals generally align with the International Committee of Medical Journal Editors’ Recommendations (https://www.icmje.org/icmje‑recommendations.pdf)—including the declaration of AI use and accountability—there are sometimes variations, as described by Koulaouzidis et al.¹ This is an exaggeration, but some journal editors may provide a policy such as, “We never permit AI use,” while others might take the opposite stance: “We have no regulation on AI use, only that you declare all.” As stated by Koulaouzidis et al,¹ AI regulation in medical writing has not yet been established, and some journal‑level freedom in this domain is permissible. However, AI guidelines are often lengthy and written flatly, making it difficult for authors to identify journal‑specific rules.

Having written over 600 PubMed‑indexed papers during my 46 years of clinical practice, I have experienced how nonadherence to journal‑specific guidelines—such as using Arial 10 pt font instead of my usual Times New Roman 12 pt font—often caused paper returns. I previously proposed that journal‑specific regulations should be highlighted on the first page of the guidelines.⁵ This applies to AI guidelines as well. If a journal has a specific AI guideline, it should be placed prominently on the first page—not embedded within general instructions or AI guidelines. If the specific guidelines are long, a simple note such as, “This journal has some specific AI regulations; refer to the corresponding section (link),” would help. If there are no specific regulations, please state: “AI regulation accords with X (publisher name); there are no journal‑specific regulations. See author guideline (link).” Otherwise, since journals have various regulations—and since those regulations could change at any time—authors must spend a considerable amount of time confirming each guideline and checking whether it has changed from one submission to the next. This burdens authors. Thus, my humble plea is: journals should minimize this extra effort by clearly highlighting any journal‑specific AI regulations.

The suggestions made here aim to help clarify important aspects of AI use—whether it is specialty‑specific or general, writing‑focused or applied in nonwriting contexts, focusing on current or future perspectives—and to urge journals to prominently highlight any specific AI‑related regulations. Such steps would provide much‑needed clarity and guidance for both authors and editors.