Perito Grid

Why journals can't find peer reviewers, and what fixes it

Paying reviewers and checking methods with AI makes peer review faster and more reliable. Here is the evidence.

By Philip Croon ·

By Philip Croon, MD, cardiology resident at Amsterdam UMC, adjunct faculty at Yale’s cardiovascular data science lab, and co-founder and CSO of Perito Grid.

Short answer: peer review runs on unpaid work, and fewer experts are willing to do it. Only 22% of review invitations were accepted in 2024, down from 43% in 2018. Journals that pay their reviewers get answers in days instead of weeks, and AI can now take over the routine checks of methods, statistics and code.

How big is the peer reviewer shortage?

Editors now send about 4.5 invitations for every review that gets done. Across 8 years of ScholarOne data, the share of accepted invitations halved, from 43% in 2018 to 22% in 2024 (Silverchair, Future of Peer Review, 2026).

The work is also unevenly spread. Researchers spent more than 100 million unpaid hours on peer review in 2020 (Aczel, Szaszi and Holcombe, 2021), and a small group of experts receives most of the requests.

The result: authors wait months for a first decision, and editors spend their time chasing reviewers instead of judging papers.

Does paying peer reviewers work?

Yes. When the journal Biology Open paid reviewers £220 (about €255) per manuscript:

  • 67% of invited reviewers said yes, compared with 23% when unpaid.
  • The time to a first decision fell from 37.7 to 5.5 working days, about 85% faster.
  • Acceptance rates stayed the same, so faster did not mean softer.

The journal has since expanded the program (Gorelick and Clark, bioRxiv 2025; Nature, July 2026).

Can AI check the methods of a paper?

For the checkable parts, yes. In a study where 45 scientists rated reviews of 82 Nature-family papers, AI reviewers did as well as or better than the best human reviewer, and added the most on rigour and code (Kim et al., 2026). AI is less suited to judging novelty or a paper’s importance to its field. That remains a job for human experts.

How Perito Grid combines both

Perito Grid is a peer review platform for academic journals. It works in three steps:

  1. The journal posts the manuscript, a link to the code and a deadline.
  2. Fisher, our AI reviewer, reviews the methods and statistics. It compares the paper with the code, recalculates the numbers, checks the references and reruns the analysis when the code is shared. Papers with serious problems can be turned down before reviewers spend time on them.
  3. A paid expert reviews the paper. Reviewers are matched on their own publications and receive €200 per review. Researchers can sign up as a reviewer.

Our AI does the checking. Experts do the judging. Diamond open access journals use Perito Grid for free, with volunteer reviewers.

Frequently asked questions

What is the acceptance rate of peer review invitations?

About 22% in 2024, down from 43% in 2018, based on ScholarOne data.

How much should a journal pay a peer reviewer?

In the Biology Open pilot, reviewers were paid £220 (about €255) per manuscript. Perito Grid pays reviewers €200 per review.

Will AI replace peer reviewers?

No. AI is good at checking statistics, code and references. Judging whether a paper matters, and deciding whether to publish it, stays with human experts and editors.

What is Fisher?

Fisher is Perito Grid’s AI methods reviewer. It checks whether a paper’s code matches its methods, whether the numbers add up and whether the references exist.

Perito Grid launches on 1 January 2027. Editors can see how Perito Grid works for journals or book a demo.

Sources

← All posts