Why Young Medical Researchers Must Struggle! JAMA Warns Against Outsourcing Writing to AI

Academic Pressure and the Commoditization of Writing

Resorting to AI is highly tempting because academic writing has become commoditized. Publications are easier to count than to read, and their impact is frequently gauged through citation metrics and media mentions. The faster researchers can produce these papers, the better.

Under overwhelming pressure from clinical and teaching demands, intense competition for grants, and concerns about academic promotion, writing a paper can feel like merely checking a task off the list. Fellows and early-career faculty quickly adopt this attitude from mentors, prioritizing efficiency over reflection or precision as their primary goal.

Research is a Creative Act of Choice

Amid these pressures, researchers can easily forget that research is fundamentally a creative act. A research paper is a formalized narrative that shows how specific choices led to specific results in a specific scientific context.

Any research study is the result of many conscious choices: selecting a question, choosing a conceptual framework, deciding on data elements, and determining a statistical model to interpret results. These choices require trade-offs dictated by constraints like time and funding. In our papers, we must disclose these choices and limitations; the precision and completeness of the writing help other researchers understand the findings so they can better replicate, extend, or refute them.

Writer Ted Chiang argues that generative AI is a "fundamentally dehumanizing technology" because it lowers expectations of both the material we read and of ourselves when we write. Chiang elaborates that meaning is achieved through a process of making choices, noting that a ten-thousand-word story requires on the order of ten thousand choices, while supplying a one-hundred-word AI prompt involves only about a hundred choices. The meaning of research is made by grappling with the "crosscurrents, backflows, and eddies" that our papers disclose, turbulence which AI may disguise as smooth, "laminar flow".

The Cost of Losing Precision

The value of human deliberation over AI speed was highlighted in a seminar focused on abstract writing. Six faculty members collectively reduced a background section from 73 words to 46 words over about five minutes. An AI program, when asked to shorten the original text, produced a 48-word version "almost instantaneously".

Despite the speed, a close reading showed that the AI version lost subtleties of language and scientific context. For instance, the AI asserted the problem was "significant," a term often reserved in scientific writing for its statistical connotation. It substituted common usage words for the specific, intentional terminology chosen by the author's community of researchers. These shortcuts resulted in less precise language, failing to accurately express the language of science in the field. Sloppily written articles reduce confidence in the authors’ thought processes, research techniques, and findings.

Final Thoughts

Ultimately, writing a scientific paper, like other creative endeavors, is fundamentally a struggle to describe the choices that articulate meaning. Efforts to improve the quality of scientific writing are not distractions from research. Instead, skillful writing exemplifies the creative process of planning, action, reflection, and communication that remains at the heart of the scientific enterprise. Young researchers may bravely consider enduring this vital struggle by avoiding relying too heavily on AI tools.

Reference: Steiner JF. Scientific Writing in the Age of Artificial Intelligence. JAMA Intern Med. Published online November 17, 2025. doi:10.1001/jamainternmed.2025.6078