KMi researchers have secured £106,078 from UKRI’s Metascience Unit to explore how large language models can streamline grant peer review without diluting fairness. The 12-month project, led by Prof Petr Knoth and Dr Francesco Osborne in partnership with Sheffield and Salford universities, will test four AI roles: triaging low-quality proposals, acting as a third reviewer, synthesising opinions as a meta-reviewer, and deploying specialist “agents” to score criteria such as originality or methodological rigour. Rare access to funded and unfunded UKRI applications lets the team simulate real panels and measure impact on decision consistency, turnaround time and reviewer workload.

By October 2026 they aim to deliver an open, transparent blueprint that UK funders—and beyond—could adopt to safeguard review quality while scaling capacity. Knoth notes that reviewer fatigue often creates lottery-like outcomes; Osborne stresses the work will keep humans firmly in charge, using AI only as a supportive, explainable tool. The award underlines KMi’s long-standing leadership in AI, scholarly analytics and ethical data science, and adds a fresh chapter to its mission of turning digital innovation into practical academic advantage.

Related links:

OU gets £106k from UKRI for a project to assess AI-enhanced grant peer review