AI Peer Review Guide
How artificial intelligence is changing academic peer review: what AI reviewer tools do, how journals use them, and what this means for how you prepare your manuscript.
AI is entering the peer review process at multiple levels: journals use AI to screen manuscripts before desk review, AI tools assist reviewers, and AI-generated reviews are an emerging concern for editors. For researchers, understanding how AI is used in peer review helps explain what happens to your manuscript after submission and how to prepare effectively. This guide covers the current state of AI in peer review and practical implications for authors.
How journals use AI before peer review
Many journals now use automated screening tools before manuscripts reach human editors. These systems check for: plagiarism (Turnitin, iThenticate), reference integrity, image manipulation, statistical errors, and basic formatting compliance. Manuscripts failing automated screens may be desk-rejected before a human editor reads them. CheckMyManuscript validates the compliance elements these systems check, reducing the risk of automated rejection.
AI tools that assist peer reviewers
Journals and publishers are deploying AI tools that help reviewers write more structured, thorough reviews. Springer Nature's 'Research Squared' summarizes manuscripts. Elsevier provides reviewer suggestion tools. Some platforms (StatReviewer) automatically check statistical reporting. Reviewers using these tools may be more systematic in identifying specific issues, which means your paper needs to meet standards across the board.
AI-generated peer reviews: a growing concern
Journals have detected cases where reviewers used ChatGPT or other AI to generate reviews without disclosure. Most journals explicitly prohibit AI-generated reviews. For authors, this is a concern because AI-generated reviews may include inaccuracies or miss key issues. If you receive a review that seems generic or contains factual errors about your paper, it may warrant reporting to the editor.
Statistical and methodological AI checks
A growing number of journals (particularly in psychology, medicine, and economics) use automated statistical checking tools at submission: StatCheck verifies that test statistics and p-values are correctly calculated; GRIM checks that means are plausible given sample sizes; SPRITE identifies impossible distributions. These tools catch calculation errors that human reviewers miss. Ensure your statistics are correctly computed and reported.
How to prepare your manuscript for AI-assisted review
Given AI's increasing role in peer review, preparing your manuscript systematically is more important than ever:
- Validate all statistical reporting
- exact p-values, effect sizes, confidence intervals
- Ensure citation completeness
- every in-text citation in the bibliography
- Complete all required declarations
- ethics, COI, data availability, author contributions
- Follow reporting guidelines
- CONSORT, STROBE, PRISMA as applicable
- Check for image integrity
- all figures clearly labeled, no misleading manipulations
- Run a pre-submission compliance check (CheckMyManuscript) to catch systematic issues
Frequently asked questions
Yes: virtually all major journals use automated tools to check for plagiarism, reference integrity, and basic formatting before human review. Publishers like Elsevier, Springer, and Wiley use AI screening at submission. Some journals also run statistical verification tools.