Queen Mary invites staff to a wider EduMark AI assessment pilot

Updated Sep 07, 2026

Queen Mary University of London announced a wider EduMark AI pilot on 15 April 2026, inviting academics across the institution to express interest. Its announcement describes an educator-controlled platform and says staff review and approve marks and feedback before release. It does not establish that every school or module had adopted the tool. Pilot announcement.

What the announcement supports

Queen Mary reports an approximate 60% reduction in marking time in earlier pilot work and encouraging student comments about clarity, specificity and usefulness. The notice does not provide a sample size, time-measurement protocol, comparator or student evaluation instrument. Treat this as a preliminary institutional claim, not an independently verified effect or a transferable saving.

A 16 February update reports $2,000 in Google Cloud credits for development, including intended work on deployment, data handling and analytics. Support through that programme is not evidence that Google audited the tool’s educational impact or security. Development announcement.

How to evaluate a local pilot

Our practical suggestion is to define both workload and educational measures before introducing a tool. Include staff time spent checking and revising draft feedback, not just generation time. Document task differences and the basis for any comparison.

Ask students whether they understand the feedback, can act on it and know how to request clarification. Check the accuracy of feedback against the work and criteria. Positive comments about presentation should not be treated as proof of fair marking or improved learning.

Agree what information enters the system, who can access it, and how staff retain responsibility. Published statements about oversight and intended secure handling need to be checked against the actual workflow before institutional adoption.

The student comment analysis governance checklist can help plan review of pilot comments. Comparing comments before and after a change can describe experiences, but cannot by itself establish the effect of the AI system.

Review note, 7 September 2026: verified the pilot and earlier institutional notices, qualified the preliminary 60% claim and distinguished development support from independent validation or university-wide adoption.