The integration of AIRA — Artificial Intelligence Review Assistant — into Frontiers’ digital peer-review platform enables better, more efficient quality control and manuscript handling
Frontiers peer review now incorporates powerful AI technology to safeguard both manuscript and peer-review quality more efficiently and keep pace with ever-growing submissions. AIRA assists editors, reviewers and internal teams by analyzing, interpreting and communicating the quality of submitted manuscripts and the review process, as well as suggesting actions and identifying potential reviewers. Built in-house and fully integrated into the Frontiers Review Forum and internal workflows, these groundbreaking capabilities have already streamlined Frontiers’ publishing process — and will continually drive further optimization through ongoing learning and inclusion of new quality checks.
“AIRA is the next generation of peer review, in which Artificial Intelligence and machine learning enable fast, rigorous and efficient peer review at ever-larger scales,” says Frontiers CEO and Co-Founder Kamila Markram. “The technology does not replace people — to the contrary, it empowers people to take editorial decisions in a more effective way, by helping our editors, reviewers and internal teams to focus on the right things and take critical decisions at the right time. This speeds up the review process and reduces time to publication, while ensuring the highest quality control.”
High-quality peer review is fundamental to scholarly publishing, certifying the validity, rigour and correctness of research hypotheses, methods and conclusions. The challenge for publishers and reviewers alike is maintaining rigorous quality standards in the face of ever-increasing manuscript submissions.
AIRA currently addresses two key peer review tasks: quality control and reviewer identification. Its algorithms quickly and accurately evaluate submitted manuscripts against a set of quality measures, including text overlap, language, the presence of human images and other ethical considerations. Those meeting a defined quality threshold are passed to editorial assignment, while any with potential issues are flagged to Frontiers’ Research Integrity and Review Operations Teams for further investigation — and will not go into review until and if resolved.
AIRA’s algorithms also identify potential reviewers based on expertise and availability, and check for potential conflicts of interests between editors, reviewers and authors — with further checks being converted from manual to AI-driven to evaluate the quality of the review itself. These are major steps forward in ensuring independent review of all manuscripts by the right experts in a field.
“AIRA allows us to evaluate submissions more rapidly, safeguard and enhance quality more efficiently and better support our authors, editors and reviewers,” says Marie Soulière, Senior Manager Peer Review, Quality and Ethics at Frontiers. “It not only flags manuscripts with potential issues, but tells us exactly what we need to manually check — such as a human image or appropriate consent. This reduces review fatigue and makes the review more accurate. AIRA also alerts us to checks that cannot be done manually, such as for plagiarism.”
The system was built in-house from custom algorithms as well as industry-leading tools from, for example, Google, iThenticate by CrossRef and Ada by Editage. All have been trained and rigorously tested using Frontiers’ 10 years of peer review experience and are fully integrated into Frontiers’ Collaborative Review Forum and internal processes.
AIRA is the culmination of 10 years of publishing experience and innovation by Frontiers. An early Open Access pioneer, Frontiers was the first to take scientific publishing entirely online through a suite of powerful, digital tools — all built and maintained in-house. These include an award-winning Open Science platform that makes high-quality, peer-reviewed research articles freely available to everyone around the world; a collaborative peer-review platform allowing real-time interactions between editors, reviewers and authors; and a virtual editorial office for editors to enhance efficiency, oversight and quality control.
Frontiers also pioneered alternative author and article metrics that provide objective ways to evaluate research impact, and launched the first social network fully integrated into a publishing platform, Loop, to maximize author and article discoverability.
“Continued innovation allows us to offer improved services to authors, editors and reviewers, and maintain a peer review process that is collaborative, transparent, efficient and rigorous,” says Markram. “AIRA takes our Open Access model to the next level of quality assurance — and will only get more powerful as we launch the many new features currently under development.”