Peer review is overwhelmed—can it survive in the AI era?
As research and AI-assisted papers surge, volunteer reviewers struggle to keep up.
Peer review was never a single reviewer's opinion. It is a distributed labor pipeline, and the HPV paper anecdote is simply a visible crack in it. Anonymous volunteers, mostly unpaid and often unanswerable, are being asked to absorb a rising tide of AI-assisted manuscripts while the scholarly community pretends the old rules still apply.
The economics of scientific publishing point the other way. Journals sell prestige and access at high margins, while the quality control that sustains that prestige is donated. That arrangement was already straining before generative AI lowered the cost of producing plausible prose, fabricated data, and hollow citations. AI tools make submission cheap for everyone, including the careless, the desperate, and the fraudulent.
The reviewer who muddled Semprini's central claim was not simply sloppy. Mistaking a policy question for a vaccine-safety question suggests someone stretched beyond their competence, and it reveals how a system that rewards decisive verdicts over careful reading can quietly damage sound research. When one mistaken interpretation blocks a paper, that is not an isolated human error; it is a structural feature of unpaid, unreviewed reviewing.
Calls for AI to rescue peer review are tempting but miss the deeper problem. AI can triage manuscripts and flag obvious statistical inconsistencies, but if reviewers are still unpaid, uncredited, and overloaded, the tool becomes another accelerant instead of a correction. The system's survival depends less on better models than on restructuring who does the work and how that work is recognized.
The labor dimension is worth watching closely. Reviewers are a volunteer workforce subsidizing a profitable publishing industry, and the pressure on that workforce is not evenly distributed. Junior researchers carry a growing share of reviews while receiving little formal credit, and seasoned scholars increasingly decline invitations. That is a market failure wearing academic robes.
The likely future is a stratified one: prestigious journals will invest in formal vetting and paid editorial oversight, while lower-tier venues lean on automated screening and hope for the best. Over time, that may widen the gap between highly guarded research and the rest of the literature. But Semprini's experience shows that even the best-staffed journals still depend on people who are neither trained nor protected. Generous assumptions about their goodwill cannot hold forever.