Elsevier submissions jumped 56% to 4.2M. Its unpaid reviewers can't keep up
Three years, 1.5 million extra papers, and a volunteer workforce already stretched thin - which is why the question now on the table is whether AI can save peer review from AI.
Elsevier, the academic publishing giant, saw submissions climb 56% in three years, from 2.7 million in 2022 to 4.2 million in 2025, while its peer reviewers work unpaid and were already stretched thin. For any leader whose decisions lean on published science, the question is whether AI can absorb the load that AI-era volume has dumped on a volunteer system.
Elsevier, the publishing giant behind thousands of academic journals, took in 2.7 million submissions in 2022. By 2025 that number had reached 4.2 million. That is a 56% jump in three years, which works out to roughly 1.5 million additional papers landing on desks that were already full. The people those papers land on are peer reviewers: academics who, by long-standing convention, read, critique, and vouch for other researchers' work without being paid for it. The source's framing puts the squeeze in a single line: reviewers were already stretched thin before the flood arrived.
That is why the question now being asked, bluntly, is whether AI can save peer review from AI. The tension is built into the phrasing. Generative AI tools have made it cheap and fast to draft text at volume, and the same technology is being pitched as the filter that could triage, summarize, and pre-screen the incoming pile. Whether AI is actually inflating submission counts is a live and contested argument. What is not in dispute is the arithmetic: more submissions, the same pool of volunteer hours.
Peer review is the load-bearing wall of the scientific record. Before a paper appears in a reputable journal, it is typically sent to two or three independent experts in the field, who assess whether the methods hold up, whether the claims match the evidence, and whether the work adds anything. Those reports decide what enters the literature and what quietly dies in the queue. Governments cite that literature when they write rules. Drug developers cite it when they design trials. Investors cite it when they underwrite a biotech thesis. The entire apparatus rests on a norm rather than a contract: senior researchers review for free, historically as part of academic citizenship, and journals bank on that goodwill.
Take the conventional two-to-three-reviewer norm and run the numbers as a rough illustration. At 4.2 million submissions, that implies somewhere around eight to twelve million review requests in a single year at Elsevier alone, up from roughly five to eight million in 2022. Those requests go to a finite set of qualified specialists, the same people who run labs, teach, sit on grant panels, and write their own papers. Adding 1.5 million papers does not add reviewers. It adds work to the people already doing it, and it adds waiting time to every author in the queue.
The volume problem has roots that predate any chatbot. Academic careers are built on publication counts, and hiring, tenure, and grant committees have historically rewarded quantity as a proxy for output. The result is a system where the incentive to submit is strong, the incentive to review is weak, and the cost of a slow review falls on the author rather than on the journal. That imbalance is structural, and structural imbalances are exactly the kind of thing automation tends to get pointed at.
Pointing AI at peer review is not a neutral fix. A model can be used to check statistical consistency, flag duplicated text, surface missing citations, and rank submissions by likely relevance, all in seconds. What it cannot do is take responsibility. Peer review's value is not the reading; it is the accountable human signature at the end of it. An editor who leans on a model still carries the reputational and legal exposure if a paper turns out to be wrong, fabricated, or plagiarized. That asymmetry, fast machine triage against slow human accountability, is the real design constraint, and it is the one nobody has solved.
The second-order effects land well beyond the academy. Corporate R&D teams, hospital systems, and regulators all treat the published record as an input. If review quality thins under volume pressure, the failure mode is not dramatic. It is a slow drift toward more retractions, more irreproducible results, and more decisions made on evidence that was never properly checked. That cost shows up years later, in failed trials, misallocated capital, and public trust that is far harder to rebuild than it was to spend.
For leaders, the practical move is to stop treating the health of peer review as somebody else's infrastructure problem. If your roadmap, your diligence, or your clinical strategy leans on published science, the reliability of the filter is a direct input to your risk. Ask the researchers you work with how many review requests they are declining, and how long their own papers sit in the queue. The headline asks whether AI can save peer review from AI. The honest answer is that AI can absorb volume, but only people can absorb responsibility, and the gap between those two is where the next decade of scientific credibility gets decided.
This story's Key Insights and Take-aways are locked.
Create a free account to unlock Executive Actions for one credit.
Register to UnlockAlways free for Executives Club members. Join the Club
More in Science
Anthropic's I.P.O. Made Rank-and-File Researchers Worth Tens of Millions
Researchers who joined just a few years ago are now worth tens of millions, and the windfall resets the pay benchmark for every private A.I. lab still paying in paper.
Sea Lion Culling Strategy Fails: New Model Says Killing Them Won't Save Salmon
The latest research from AAAS overturn a long-held conservation assumption, forcing policymakers to rethink critical resource allocation in marine ecosystems.
Microplastics do more than pile up: they react with water to damage cells
Scientists have found that plastic particles can generate highly reactive molecules in water, a mechanism that could explain links to heart attacks, strokes, and neurological disorders.


