For anyone who participates in the publishing process – authors, reviewers, and editors – it can be a frustrating and tedious process.
For authors, the process begins with formatting and submission. Virtually all major journals (and publishers) have their own unique format; hence, when an author has a manuscript rejected and submits it to a different publisher, it often must be reformatted. Submission is challenging and time-consuming, not just uploading the manuscript and various requested forms, but checking dozens of boxes – only to have the submission fail quality control and not be “accepted” for review until all information is 100% correct. Then, of course, is the long wait for a decision – thankfully, rejection without review usually comes quickly, but if a paper is sent out for external peer review, the wait is often months, sometimes many months!
For peer reviewers, there are relentless requests to review. If a reviewer accepts an invitation, they must sign on to a specific manuscript-processing system, claim and read the manuscript, write a review, and then jump through more hoops, checking more boxes, to submit it.
For editors, it is the incessant increase in the number of submitted manuscripts, the interaction with often archaic manuscript management platforms , and the challenge of identifying peer reviewers. Then, a decision needs to be made. If a revision is requested, it is necessary to assess whether the authors did what they say they did.
Even out of this morass come many important articles published in great journals.
We envision a very different future – with AI assisting in submission, reviewing, and revision, with the possibility of going from submission to publication within a few weeks rather than months. Just as AI is being used to help clinicians -- writing progress and discharge summaries, or ambulatory notes -- AI will assist authors with formatting and submission, peer reviewers (if they are used at all) with review, and editors with review and revision consideration.
Let’s just consider the review process itself.
Many have said the current process is flawed—biased, inefficient, and often ineffective—but external peer review is nevertheless indispensable. That is now a testable hypothesis. Can external peer review by humans be eliminated from the process, producing a less biased and more efficient process, with submission to publication occurring in just weeks, largely dependent on how quickly an author can revise a manuscript?
This is how it would work. A manuscript would be submitted for consideration; an initial AI review would be presented to an editor with a score, based upon rigor, originality, and importance, and the editor would decide to reject or revise, with the AI review returned to the author. For those manuscripts requiring revision, the revision would undergo a 2nd AI review. If the score developed by AI improved substantially , the manuscript would be published, along with the 2 AI reviews and the initial submission.
Time from submission to publication – days to weeks. No external peer review would be used. Domain experts could assess originality and importance at any step along the process if it is deemed important.
AI could assess aspects of review that currently depend on humans. For randomized clinical trials, was the trial registered, and if so, did the authors adhere to trial registration and the related reporting guideline – CONSORT. AI could assess adherence to any of the hundreds of reporting guidelines, check for fabricated references, detect AI use in preparing the manuscript, and verify whether appropriate statistical tests were used. Editorial review would include only reading the abstract to ensure the conclusions could not harm patients or public health. For example, any article that “reports” a relationship between immunizations and vaccines, or that PrEP is not effective, would have to undergo human peer review.
This description is obviously an idealized view of how AI can and will change scientific publishing. Some of these changes are already happening. Some journals are using AI as an assistant; peer reviewers, regardless of instructions from publishers, are using AI; and authors are using it for many aspects of research.
Of course there are challenges. It is imperative to “protect” the data submitted by authors. AI is likely to be biased, but the important question is whether it is more or less biased than humans, and once that bias is noted, can it be eliminated? AI review often includes dozens of recommendations, some important, others less so, or even controversial. Can models be trained to focus on issues, such as methodological and statistical rigor? Most importantly, can AI assess originality and importance – or is that truly a human enterprise? Given that humans are constrained by what they can remember, while AI has access to 30 million publications in PMC – it may be better at evaluating originality.
Over the past 30 years, there have been several disruptive events in medical publishing. The introduction of the internet altered how scientific information is communicated – moving from print to electronic – with major journals becoming international communication networks. Open access publishing – where the fundamental relationship between author and publisher changes- altered the business model.1[i] Mandated registration of RCTs, often with a protocol and statistical analysis plan, has substantially improved the integrity and reporting of RCTs. Preprints, although not as disruptive or important as the others, were initially met with skepticism in clinical medicine. They have now largely become acceptable for most publishers.
AI is the next great disrupter – and like the Internet – will radically alter medical publishing.
Howard Bauchner, MD, is a Professor of Pediatrics and Public Health at the Boston University Chobanian & Avedisian School of Medicine. He is also a Visiting Scholar at the National University of Singapore. Dr. Bauchner is also the former Editor-in-Chief of JAMA and the JAMA Network (2011-2021) and Archives of Disease in Childhood (2004-2011).
Frederick P Rivara, MD, MPH, is Professor of Pediatrics and Adjunct Professor of Epidemiology at the University of Washington. He is also the former Editor-in-Chief of JAMA Pediatrics (2000-2017) and JAMA Network Open (2018-2024). He continues as an active clinician, mentor, and investigator.
Open access has had many unintended consequences, including unscrupulous publishers, huge profits for the traditional publishers, and making millions of articles available to large language models for consumption.





Dr. Bauchner and Dr. Rivara describe the frustrations I have watched keep good clinical cases from ever being published: reformatting for each journal, dozens of boxes to check, and months of waiting only to be rejected. That process is reason enough never to write the case up, and those cases are lost.
I built Rural Hospitalist Case Reports to take a different approach. Physicians share their cases directly. There are no formatting requirements, no word limits, no fees, no rejection, and no waiting. A case is published when the author decides it is ready, under a CC-BY license with a permanent URL. We are also working on assigning DOIs so each case can be formally cited. The idea is that the clinicians who see these cases are in the best position to judge whether they are worth sharing, and readers can judge the rest.
The site is new, rudimentary, and very much a work in progress, so I welcome feedback from anyone willing to try it or share a case:
https://rural-hospitalist-case-reports.web.app
Kirk Dodson, MD