Will the day come when opinion and letter pieces disappear from the medical journals?

Abstract

Here, I claim: (1) individual writing tone holds value in medical papers, especially opinions and letters; (2) Generative artificial intelligence (AI), like ChatGPT, may erode this tone; (3) ChatGPT will produce abundant opinions/letters with similar tones; and (4) journals, inundated with such submissions, might stop publishing these sections. Here, I use ChatGPT to represent generative AI. ChatGPT is already widely used in medical writing. Having reviewed over 2600 papers (Web of Science record), I have recently noticed a decline in “very poorly written” manuscripts: many authors now rely on ChatGPT, either partially or heavily. Last month, a professor said to me, “I reviewed a paper. I believe you were the other reviewer, right?” Right! The paper was accepted, and this exchange raises no ethical concerns. Recognizing my writing style, the professor was confident it was me. After 40 years of writing, some can identify me by my style. Through years of effort, I have developed a distinctive writing tone.1, 2 In medical papers, “what's written” comes first, but “how it's written” also matters. Just as individual “diversity” is vital for a “healthy” society, diversity in writing may hold meaning for “healthy” medical journals. Whether one likes my tone is another matter. For a humble experiment, I input the essence of points (1)–(4) and tasked ChatGPT-4 with generating a short opinion/letter (Table 1). I'm unsure which is “better.” However, for those who do not value “personal tone,” the two Letters are essentially the same, as “what's written” is identical. Input (prompt) I expect an unsettling future image of medical writing and journals. The points are: (1) individual writing tone holds value in medical papers, especially opinions/letters; (2) ChatGPT may erode this tone; (3) ChatGPT will produce abundant opinions/letters with similar tones; and (4) many such ones will be submitted, and thus journals might stop publishing these sections. Supporting anecdote and data: (i) Some colleagues of mine grasped my writing tone and recognized that the review sheet was written by me. (ii) Data showed that AI-generated “poems” were evaluated better than human-generated ones. Task: Involving points (1)–(4) as well as (i) and (ii), generate a short letter claiming that the use of AI in medical writing will erode the future of the journal world. At present, we do not know what to do. ChatGPT-generated letter Title: The erosion of individuality in medical writing: A perspective on AI use Dear Editor, Medical writing, particularly in sections like opinions and letters, thrives on individuality. Each author's unique tone, shaped by their experiences and insights, contributes to the richness and diversity of the literature. Anecdotally, colleagues have remarked on recognizing my writing tone in anonymous review sheets—a testament to the distinctiveness that seasoned authors cultivate. However, the rise of generative AI tools like ChatGPT threatens to erode this individuality. AI-generated text, while efficient, lacks the nuanced human touch. It homogenizes tone, producing content that is polished yet devoid of personality. As a result, we may face a deluge of AI-crafted opinions and letters that are strikingly similar in style, diluting the diversity that once defined these sections. Evidence from other domains, such as poetry, supports this concern; studies have shown that AI-generated poems are often rated higher than human-written ones, suggesting a preference for uniformity over authenticity. This trend poses a significant risk to the journal ecosystem. A flood of AI-generated submissions could overwhelm editorial boards, leading some journals to reconsider publishing these sections altogether. Losing the platform for genuine human expression would be a disservice to the medical community, which relies on diverse perspectives to foster critical dialogue. At present, the path forward remains uncertain. Balancing the efficiencies of AI with the preservation of individual writing tones is a challenge we must address to safeguard the integrity of medical writing and its journals. Opinions/letters typically require less evidence but more personal perspectives and thoughts than original articles.2, 3 They allow, and indeed require, authors to freely express their ideas. If editors deem them useful to the medical community, they are accepted; otherwise, they are not. I believe well-crafted opinions/letters, both in concept and tone, add a unique “flavor” to a journal.2, 3 ChatGPT generated the Letter in just 1 min. This demonstrates that anyone who understands points (1)–(4) can produce such a Letter. Journals may soon be inundated with submissions of “linguistically perfect” yet “homogenous and monotonous” Opinions/letters, potentially forcing them to discontinue this section. AI has even created “poems” that were rated superior to human-written ones.4 Poetry, a genre that relies heavily on individual “diversity,” has been successfully generated by AI. So, it is unsurprising it can handle opinions/letters. I deeply enjoy reading and writing opinions/letters. I can only hope the future proves me wrong. Shigeki Matsubara: Conceptualization; investigation; writing – original draft; writing – review and editing. A part of the present concept was published and is cited appropriately. The author has no conflict of interest. I used ChatGPT-4 to generate the Letter (ChatGPT-generated Letter; Table 1). Data sharing is not applicable to this article as no new data were created or analyzed in this study.

Ähnliche Arbeiten