Clarifying contexts and audiences

Professor Matsubara calls for clearer distinctions in discussions about AI—between specialty‑specific and general use, writing and nonwriting contexts, and present realities vs future possibilities. We share his concern that with a rapid spread of tools such as ChatGPT, broad generalizations pose a risk of obscuring important domain‑specific implications. In our original letter, we adopted a deliberately broad lens, reflecting the cross‑disciplinary nature of AI’s impact on medical communication. However, we recognize that readers would benefit from more structured segmentation of the issues.

While our letter referenced the influence of AI in diverse publishing environments, from high‑impact journals to more practice‑oriented periodicals, we acknowledge that delineating specialty relevance could enhance interpretability for our varied audience, which includes clinicians, researchers, AI developers, and editors. Likewise, distinguishing between “writing‑focused” and “nonwriting” use cases is essential. AI tools used for data analysis, diagnostic support, or clinical decision‑making operate under different constraints and expectations than those used for manuscript drafting or literature summarizing. We welcome this framing and will strive to adopt this taxonomy more explicitly in future contributions.

Current limitations and future horizons

Professor Matsubara’s suggestion to distinguish between the “current” and the “future” states of AI capabilities is both practical and necessary. As he points out, while today’s generative AI models are prone to hallucinations, biases, and contextually incorrect outputs, tomorrow’s models will be more refined, potentially even integrated with citation‑based accuracy mechanisms and contextual understanding beyond the sentence level. At the same time, he raises a valid concern about how AI might affect human thinking, especially our ability to write well, think originally, and make sound ethical judgments. We share his concern. Leaning too heavily on AI poses a risk of weakening the very skills that drive scientific reasoning and thoughtful argument. Thus, our position remains unchanged: AI should serve to augment, not replace, the reflective, ethical, and creative faculties of human scholars.

Journal‑specific guidelines

One of Professor Matsubara’s most practical recommendations pertains to the editorial guidelines of scientific journals. We strongly support his call for increased transparency and visibility of AI‑related policies. As he noted, variations between journals and even between publisher‑imposed standards and journal‑level specifics often require authors to decipher lengthy documents that do not clearly state the permissible scope of AI use.

This issue is all the more pressing as disclosure requirements grow increasingly complex. Professor Matsubara’s idea—a clear, standard statement right at the top of journal instructions —just makes sense. It would save authors time and cut down on mistakes. His example of manuscripts being bounced back for small formatting errors is a good reminder: little things can block big work. The same applies now to AI use. Misreading or mishandling these new rules could lead to the same kinds of problems. That is why we think journals should consider a tiered system for disclosure policies, such as: 1) Tier 1: a summary statement on the first page of the author guidelines, indicating the presence or absence of AI‑specific rules; 2) Tier 2: a link to detailed guidance, including examples of acceptable and unacceptable uses (eg, grammar correction vs content generation); 3) Tier 3: a requirement for authors to disclose any AI usage in the manuscript submission form.

Such an approach would benefit authors, reviewers, and editors alike, enabling clarity and accountability throughout the publishing workflow.

In conclusion, we are grateful to Professor Matsubara for enriching the conversation on AI in medical publishing. His reflections serve as a reminder that responsible technological adoption is not merely a matter of capability but of context, clarity, and community standards. As AI continues to reshape academic medicine, our collective challenge will be to preserve the integrity of human insight while embracing the efficiency and potential of intelligent systems. We believe the points raised in both our article and Professor Matsubara’s letter will help shape a more thoughtful and transparent framework for AI integration into the medical sciences. We encourage continued dialogue and hope that editors, publishers, and academic bodies will take these concerns into account as they refine their approaches.