Exploring New Approaches to Assessments with Generative AI

At the University of Exeter, the Education Incubator supports innovative ideas that have the potential to transform teaching and learning. This year, I am excited to join a project that aims to reimagine the way we assess mathematics! Specifically, within the module MTH2011 Linear Algebra, by exploring how Generative AI can support both assessment design and marking.

Exploring Generative AI tools for assisting human markers

This Incubator-funded initiative, led by Professor Barrie Cooper (Mathematics & Statistics) and Pierre Camaly de Brosses (Learning Development), is investigating whether Generative AI tools can assist human markers in evaluating student work. We aim to integrate AI to reduce the workload on markers by flagging the fully correct answers, allowing human markers to focus on marking partially correct answers. This does not mean we will use only AI for marking, but we will automate the process to reduce human errors and workload, while still having human markers in the loop!

Our aim is not simply to automate processes, but to understand how AI can be used responsibly, transparently, and in partnership with educators. By evaluating both the process and outcomes, we hope to generate evidence-based insights that can shape future practice across the University.

Since the start of the project, we have been exploring a range of Application Programming Interfaces (APIs) and solution architectures to understand how AI can best support the marking of mathematical work, particularly labelled diagrams, which can be time-consuming for human markers to process consistently. Early comparisons of AI tools have already helped us identify promising strategies and areas that need further refinement.

   

Early Highlights: Presenting at AWS PartyRock

Our project team recently attended the Amazon Web Services (AWS) PartyRock event at Exeter, where I had the opportunity to present our work and discuss the possibilities of integrating Generative AI into learning and assessment. Connecting with others working on AI-driven educational innovation was incredibly inspiring, and it reinforced the relevance and importance of our project in the broader conversation about the future of education.

Why This Work Matters

Assessment plays a central role in student learning. Yet, it remains one of the most labour-intensive and challenging areas for educators, especially in mathematically complex modules. By exploring the potential of Generative AI to support bespoke assessment creation and marking, our project seeks to:

• Increase fairness by reducing opportunities for academic misconduct
• Enhance the authenticity and variety of mathematical assessments
• Support staff through more efficient and transparent marking processes
• Build confidence among both educators and students about the responsible use of AI tools

Through careful evaluation and collaboration with the mathematics community, we aim to identify which approaches are practical, what limitations exist, and how these tools can best complement the work of human markers.

Looking Ahead

From January 2026 onwards, our focus has shifted to real-world evaluation. Academic staff will begin using the AI-supported workflow in an authentic marking environment, and we will gather data on efficiency, accuracy, user experience, and the overall value of the solution. This includes conducting surveys, interviews, and focus groups with markers and students. Insights gathered during this period will shape further refinements, ensuring that the tool is not only functional but genuinely helpful.

If you are interested in this work and would like to be involved in the evaluation phase, we would love to hear from you. Your feedback will help us understand how Generative AI can support assessment marking and how similar approaches might benefit other areas across the University. You do not need to be a marker or a student in MTH2011; anyone curious about AI in education is welcome to join us in exploring and shaping this innovative approach.

Over the coming months, we will continue designing assessments, collecting data, and documenting our findings. I am excited to contribute to this innovative work and to help develop insights that may inform future teaching, learning, and assessment practices at Exeter and beyond.

If you’d like to take part or learn more, please feel free to get in touch with me at Z.Guler@exeter.ac.uk. We’d be delighted to collaborate with you! I look forward to sharing more updates as the project progresses.

 

For more information please contact: