School of Education Research Seminar

Beyond Literacy to Orchestration: Designing Valid and Sustainable Assessment in the Age of AI

Dr James Wood, Durham University

Abstract

This session moves beyond the moral panic around AI and cheating to explore how higher education can integrate AI into assessment in principled and sustainable ways. It begins by addressing the core challenge AI poses to assessment validity, critiquing detection-based approaches and emphasising the need for tasks that generate authentic evidence of learning. A framework informed by Cognitive Load and socio-constructivist theory will distinguish between valid uses of AI i.e. those that reduce unnecessary workload or provide socio-constructivist scaffolding and uses that bypass the cognitive effort essential for learning. Using a case study of process-oriented portfolios, we illustrate how assessment can be designed to make cheating “harder to fake than to make” while fostering students’ capacity for learning orchestration: strategically seeking, evaluating, and integrating feedback from teachers, peers, resources, and AI. The session concludes with practical implications for building assessment systems that uphold validity, strengthen agency, and support workload sustainability.

Biography

James Wood recently joined Durham University as an Assistant Professor in Digital Literacy and Assessment. Prior to this appointment, he held several academic leadership positions at Bangor University, including Director of Assessment, Director of Postgraduate Taught Programmes, and Director of the National EdD Programme, which he helped design and deliver. His teaching experience also spans Seoul National University, King's College London, and UCL.

James serves on the executive committee of the Assessment in Higher Education (AHE) Network and sits on the International Editorial Board of Assessment and Evaluation in Higher Education. His research examines how technology, including AI, can enhance learner agency, self-assessment, and co-regulation in feedback processes, drawing on socio-constructivist and socio-material perspectives. He is particularly interested in designing assessment and feedback practices that promote deep learning and assure quality while remaining workload-sustainable and fostering relationality and trust.

Joining Online
This event will be accessible via Microsoft Teams. If you would like to attend online, please contact ed.research@durham.ac.uk to request the Teams link.