Leiter Reports: A Philosophy Blog

News and views about philosophy, the academic profession, academic freedom, intellectual culture, and other topics. The world’s most popular philosophy blog, since 2003.

  1. David Wallace's avatar
  2. Mark Taylor's avatar
  3. Jeff Moriarty's avatar
  4. AG Tanyi's avatar
  5. Douglas W. Portmore's avatar
  6. Gregory C. Mayer's avatar
  7. Brad's avatar

Is AI destroying the peer-reviewed journal publication process?

Brian Leiter

Anna Wexler, a bioethicist and journal editor at Penn’s medical school, writes in The Atlantic:

Journals are being inundated with manuscript submissions. Some journals have reported their submissions doubling or even tripling since ChatGPT was released, a few years ago. Across the Elsevier portfolio of journals, submissions were about twice as high in 2025 than in 2024. One journal reported having to double its editors—all unpaid volunteers—to handle the boom in submissions.

Reviewing a growing number of submissions with a critical eye now takes time that minimally compensated or unpaid journal editors, including me, do not have. As a result, even as journal editors desk-reject more manuscripts, they are also sending more and more AI-assisted papers out for review, which in turn off-loads the work of picking apart AI-assisted articles to two or three other overworked scholars. These scholars are, in turn, receiving more and more review requests. In some cases, the cycle has become absurd: AI-assisted papers land with overworked editors, who send them to rushed reviewers, who in some cases use AI to write the review, which the authors revise and resubmit using AI.

A recent study of 7.3 million journal articles published from 2020 to 2025 found that by 2025, more than half showed evidence of LLM assistance, up from 12 percent in 2023. These problems are most likely to afflict the humanities and social sciences, because conceptual argumentation is easier to mimic than experimental results. But AI can generate much of an article even in the hard sciences.

Curious to hear from journal editors (or peer reviewers or recipients of peer reviews) about their experiences. Please use a real email address, and if you do not use your full name, say something about who you are (e.g., journal editor in philosophy or law [or some other field], junior faculty in [field], peer reviewer in [field], etc.).

, ,

Leave a Reply

Your email address will not be published. Required fields are marked *

5 responses to “Is AI destroying the peer-reviewed journal publication process?”

  1. Douglas W. Portmore

    Submissions to ETHICS are up this year, but not dramatically so. I project that they’ll be up around 9% for 2026. Submissions have also increased each year for the last few years. Last year, for instance, submissions were up by around 5%. I suspect this increase is largely because I’ve made submitting to ETHICS less costly (opportunity-wise) by dramatically reducing the time for us to reach an initial decision. Of course, I have identified a few papers as AI-generated (where these submissions consist mostly of a series of bullet points), but not many. Also, my (EIC) desk rejection rate has risen over the last few years, but only by a couple of percentage points each year. This is largely due to the need for me to be more selective given the increase in submissions, but it’s also partly due to my getting better at determining which papers have little chance of surviving the later review stages.

    Reply
  2. AG Tanyi

    As member of editorial boards (but not as editor), there clearly is an uptick in submissions. As reviewer, I sometimes send back submissions that are clearly mass-produced by AI (i.e. it is not a creative process that perhaps also involved some use of AI, but to my eyes anyway, someone just made the machine write the whole thing from scratch).

    A more general comment: given that AI submissions are not allowed, it is really humans, that is, our colleagues who are destroying the system, not AI per se. If anyone is destroying anything, that is.

    Reply
  3. Jeff Moriarty

    I am the co-editor-in-chief of Business Ethics Quarterly. About half of the articles we publish are normative/philosophical. The rest are empirical social science (these are handled by my colleague Danielle Warren at Rutgers).

    20 years ago, the journal received less than 150 articles per year. 10 years ago, we had just over 300 per year. Between 2020 and 2024, we averaged 400 manuscripts per year. In 2025 we received 518 submissions and in 2026 we are on track to receive well over 600. In October alone we received 63 manuscripts. All this is to say: our submissions have been going up steadily for 20 years, but AI seems to have caused a major increase.

    A lot of the new submissions seem to be AI slop, which we reject, but it takes time to figure this out. I am not under the illusion that I find it all. Opening up Manuscript Central to check my queue is usually not fun.

    This does not feel sustainable. I am eager to hear how others are dealing with this and what they think should be done.

    Reply
  4. Mark Taylor

    I’m not an editor, but I have so far been able to keep to a policy of always accepting peer review requests from journals for topics that I am qualified for. Those requests have skyrocketed recently; I received and completed four last month alone. The average quality of manuscripts is so low that I worry my standards might be eroding–I get really excited to see a manuscript that is well-written and has even a half-interesting idea (hunger is the best spice, I suppose).

    In terms of solutions, the most concrete I can suggest is that all journals ought to immediately start requiring authors to include page numbers with all of their citations. This helps in several ways: (1) AI is very bad at providing page numbers since it frequently doesn’t have access to the whole work being cited. It tends to work with abstracts, since those don’t require access permissions, which is also why its claims are frequently so vague and general. (2) Validating citational claims and hallucinated sources is much faster if there are specific page numbers. I recently had a manuscript cite all of Taylor’s “Sources of the Self” in support of a claim I was pretty sure appeared no where in that book; but proving a negative in that case is very time consuming given the book’s length. (3) It would allow editors/reviewers to start review by quickly spot-checking citations for hallucinations and sometimes disqualifying manuscripts on those grounds without needing to get into the more time-consuming work of working through the substance of the manuscript.

    It would also be great if journals communicated with each other about authors who provide demonstrably false sources. In addition to the false citations found during peer review, I have been dismayed at some new citations of my publications which are completely bogus (i.e., they cite me in support of claims I don’t even discuss). I suspect that much of this misbehavior is coming from a relatively small pool of repeat offenders. They should be barred from further self-serving attempts to bog down and pollute academic journals.

    Reply
  5. David Wallace

    If you don’t want your journal to publish papers that are written by AI (and I think you should not, though I think it’s a great research tool in other ways) just have a policy to that effect and then use Pangram to enforce it.

    Reply

Designed with WordPress