According to a single report from Ars Technica, the academic peer-review system is struggling to cope with a rising number of research submissions, some of them AI-assisted. The article describes health economist Jason Semprini, whose study on HPV vaccine mandates was rejected after a single reviewer reportedly misunderstood its central argument—confusing a study of policy effectiveness with a question about whether the vaccine itself prevents cervical cancer. Papers are often reviewed by two or three experts, but Semprini's had only one. The piece cites figures indicating that papers indexed in Scopus and Web of Science have been increasing at 5.6 percent per year, and that researchers globally devote an estimated 15,000 years of work annually to peer review, valued at $1.5 billion for the US share alone. Editors describe growing difficulty recruiting reviewers: Steven Mack, an editor for Human Immunology, said he recently emailed about thirty researchers to find one willing reviewer, compared with five to 10 emails to find three willing reviewers five years ago. The article notes that some researchers, particularly in AI, are shifting toward informal channels such as blogs. Historian Aileen Fyfe is quoted noting that universal peer review is only about 50 years old.
- Papers indexed in Scopus and Web of Science are reportedly rising 5.6 percent per year
- An editor said finding one reviewer now takes ~30 emails, versus 5-10 for three reviewers five years ago
What it means for you
The system scientists use to check each other's work before publishing is buckling under the sheer volume of new papers, and AI-assisted writing is adding to the pile. For most readers this is background context, not a call to action—but it's a useful reminder that a claim being 'published' or 'peer-reviewed' means less than it used to.
Try this
When you see a study cited to justify a product or a business claim, check whether it was actually peer-reviewed, how many reviewers looked at it, and whether independent groups have replicated it—rather than treating 'a study found' as settled fact.
Who should care
Anyone who relies on scientific findings to make decisions—healthcare-adjacent businesses, educators, journalists, and people who cite research in their own work.
Skip this if
You don't publish, review, or make decisions based on academic research; this changes nothing about your day-to-day work with AI tools.