[The Role of Artificial Intelligence in Peer Review: Pros, Precautions, and Practices]
The Scientist readers share their thoughts on the use of AI in peer review.Image credit:©iStock.com, Natalya Kosarevich
A-I tools have become increasingly common in research, covering everything from multi‑agent AI systems that automate experimental design to utilities that assist manuscript composition. As their prevalence increases, the academic community raises key questions about whether—and how—they should be incorporated into peer review processes. To gather diverse perspectives, The Scientist commissioned its readers to discuss AI’s potential role in assessing scholarly work. Below are representative insights from recent contributors.
Dr. Dougherty, who has served on the editorial board of an American Society for Microbiology journal for over two decades, asserts that even amid widespread AI adoption he has not relied on machine‑based manuscript evaluation. He points out that the significant effort authors expend does not justify depending on tools whose outputs are inherently limited. Moreover, many pivotal methodological details are concealed behind journal paywalls, and historical literature often eludes algorithmic ingestion, depriving AI of essential contextual knowledge. Consequently, he insists that peer review should retain its indispensable human component.
Walker supports AI utilization as a supportive instrument. He warns against treating AI as an autonomous final reviewer, emphasizing instead its value as an auxiliary aid that supplement existing QC practices.
Megighian proposes revisiting historical peer‑review mechanisms. He envisions an online portal where submissions can be posted freely, enabling communal, blog‑style discussion on methods, results, and overall merit, thereby restoring the interactive exchange characteristic of in‑person scientific conferences.
Fokialakis holds that AI is most valuable when leveraged as a productive helper rather than as the sole decision‑maker in the review cycle.
Marshall tends toward adopting AI as part of her workflow, citing its extensive synthesis of interdisciplinary knowledge. While she prefers not to delegate exclusive responsibility to machines, she welcomes AI to operate alongside a seasoned human reviewer to enhance comprehensiveness and efficiency.
Arijit observes a mixed impact of AI in peer review. He advises its deployment primarily as a pre‑check step that flags potential issues for professional scrutiny, cautioning that allowing AI to generate substantive critiques risks diluting the epistemic rigor that defines true peer review.
All responses have been edited for brevity and clarity.