Peer review is overwhelmed—can it survive in the AI era?
A system built on volunteer goodwill is buckling under a flood of papers, many of them written with an assist from AI.
Written by OutOfToken AI
August 10, 2026 · 4 min read · Synthesized from reporting by Ars Technica · How this works
Scientific publishing runs on an old-fashioned promise: experts read each other's work for free, catch the mistakes, and vouch for what gets published. That promise is breaking. Submission volumes are surging, AI is both fueling the flood and degrading what's inside it, and the unpaid reviewers holding the system together are running out of road.
The math was already broken
Even before generative AI entered the picture, peer review was strained past sustainable limits. At many medical journals, editors report receiving between 30 and 100 manuscripts a day, all needing qualified reviewers willing to work for free. That imbalance between reviewer supply and manuscript demand predates ChatGPT — AI just poured fuel on it.
ChatGPT opened the floodgates
Organization Science, a leading management research journal, saw submissions jump 42% after ChatGPT's late-2022 debut. Editors there didn't find a productivity boom. Instead, the journal's AI Task Force found the added papers were often harder to read, thick with jargon, and less likely to survive review.
"Submissions up 42% post-ChatGPT — but editors say the extra papers aren't better, just harder to review."
Slop at scale
AI tools have crossed a threshold: they can now generate convincing-looking papers nearly from scratch. That capability has drawn in researchers under pressure to publish, producing what critics increasingly call scientific 'slop' — manuscripts that look legitimate on the surface but strain or fail under scrutiny. The consequence lands squarely on reviewers, who must now spend more time simply determining whether a submission is worth serious evaluation at all.
Reviewers are quietly exiting
The strain isn't abstract. Researchers describe manuscripts sitting under review for months, and warn that peer review's reliance on unpaid labor means the system was at capacity long before AI accelerated output further. Those with the least institutional support — early-career researchers, scientists at smaller institutions — are often the first to walk away when the wait becomes unworkable, raising real concerns about talent leaving science not because of funding cuts, but because the pipeline itself has become a bottleneck.
AI as scalpel, not replacement
Publishers aren't ignoring the problem. AI tools are already deployed for integrity checks, language polishing, reviewer matching, and summarizing manuscripts so editors can grasp key points faster — all live in production editorial workflows today. But experts are careful to draw a line: these tools assist human judgment, they don't replace the accountability a human reviewer provides.
Peer review's core vulnerability was always its dependence on volunteer time in a system with no real capacity ceiling on submissions. AI has exposed that fragility rather than created it, accelerating both the volume of research and the volume of noise. Whether the system survives may hinge on whether reform arrives faster than reviewer burnout does.
Editorial Note
The research strongly corroborates the article's core claims about submission volume surges post-ChatGPT, the degraded quality of AI-assisted papers, reviewer burnout, and the pre-existing capacity crisis. All major factual assertions are supported by cited sources. The research does not dispute the article's framing of peer review as broken, though it offers some nuance (Source 5 notes AI tools can assist peer review) that the article downplays.
Claim Tracker
AI-assessed
Source 6 (Artificial or Intelligent? The Impact of AI on Academic Publishing) states: 'In most fields of scientific inquiry in medicine, the average peer-reviewed journal receives 30 to 100 manuscripts every day.'
Source 4 (AI is flooding peer review) confirms: 'At Organization Science, one of the leading journals in management research, submissions jumped 42% after ChatGPT arrived in late 2022.'
Source 4 corroborates: 'Many were harder to read, more stuffed with jargon, and less likely to survive the review process.'
Source 3 (The Verge) states: 'AI has improved to the point where it can produce convincing papers almost wholesale.'
Source 2 (LinkedIn post by Renée El-Gabalawy) confirms: 'Peer review is running on goodwill and unpaid labor, and that system was already at capacity before AI began accelerating research output.'
Ask AI about this story
// discussion
sign in to join the discussion