Students and lecturers come up with 32 rules for fair assessment in the age of GenAI
Can assessment practices become more fair in the age of GenAI? A group of Leiden University students and teachers set out to work on this question during the FAIR-ASSESS Deliberative Assembly. One key takeaway: our views on GenAI remain diverse – and that’s a strength of our university community, according to the FAIR-ASSESS project team.
Generative AI tools are now used across higher education in multiple ways. Assessment practices are no exception: not only do many students worldwide use GenAI when completing assessments, some teachers also use the technology for designing assessments, and some experiment with using GenAI for feedback and even grading purposes. While these developments bring opportunities, they also raise significant concerns, particularly regarding validity, transparency, and compliance with legal frameworks such as the EU AI Act and institutional examination regulations. Against this background, the FAIR-ASSESS project posed a central question: how can assessment practices become more, rather than less, fair in the age of GenAI?
Need for clear regulation
To address this question, a diverse group of 47 students and teachers from all seven Leiden faculties participated in a deliberative assembly. Over four half-day sessions spread across five weeks in January and February 2026, participants engaged in a structured process of learning, discussion, and decision-making. They received an introduction to GenAI technologies and current research on their impact on higher-education assessment, examined both potential benefits and risks of GenAI use at different stages of the assessment process, and reflected on the need for clear regulation within Leiden University.
A central outcome of the assembly, as can be read in the project team’s recently published advisory report, was the development of a set of 32 proposed rules, complemented by broader recommendations on access to and use of GenAI in assessment. The rules are grounded in principles that participants identified as essential for assessment at Leiden University, including validity, reliability, transparency, usability, trust, authenticity, responsibility, accountability, equality, and academic integrity. Protecting these principles was seen as crucial to maintaining the quality of and confidence in assessment practices that are subject to a rapidly changing technological context.
Insights and inspiration for policy development
At the same time, the voting process demonstrated that views on GenAI use remain diverse. No single rule received unanimous support, reflecting differing disciplinary traditions, experiences with technology, and normative views on assessment. The editors of the advisory report therefore recommend that the resulting set of rules should be read as a collection of guidelines that achieved moderate to substantial support, rather than as a statement of full consensus.
According to the editors, the outcomes of the Leiden Deliberative Assembly offer a valuable resource for students, teaching staff, and administrators. They provide insights and inspiration for ongoing discussion and policy development on GenAI use in assessment and can inform future regulations at Leiden University. More broadly, the assembly illustrates how inclusive, deliberative approaches can support thoughtful and balanced responses to the challenges and opportunities posed by emerging technologies in higher education, and how reflecting on these challenges and opportunities together can make our university more democratic.