In the hope of maximising the use of staff time, and facilitating the actioning of student input quickly and effectively, the AI Teaching and Learning Exploratory Fund enabled the Nuffield Department of Primary Care Health Sciences to explore whether AI could streamline the student feedback process, reduce admin burden, and enhance teaching quality.
The ‘AI-Driven Feedback Loops for Student Evaluation’ project, led by the Nuffield Department of Primary Care Health Sciences, set out to explore whether AI could help streamline and standardise feedback processes across its postgraduate courses. Prior to this project, each new programme in the department had been implementing its own feedback processes, meaning that there was no consistent administrative approach to collecting, analysing and responding to student feedback.
Over six months, the team piloted the use of ChatGPT to support feedback analysis, with the aim of reducing the staff time required to manage feedback while delivering more consistent and insightful outputs. Unfortunately, the experiment did not yield the hoped-for proof of concept. In fact, using current AI models increased rather than decreased staff workload due to the need for extensive prompt crafting, interpretation, and quality assurance.
With training and support from Kelly, Lead Business Technologist in the AI Competency Centre, staff engaged hands-on with the technology - developing valuable AI literacy, albeit not the intended outcome. Participants gained confidence using AI for other academic tasks such as summarising documents and proofreading, suggesting unexpected benefits beyond the project’s initial scope.
The project highlighted key lessons: while ChatGPT shows promise for simple tasks - like drafting survey questions or summarising already-analysed data - it is not yet reliable for primary analysis of complex, nuanced feedback. Human context and judgement remain crucial. For now, if staff choose to use AI in this space, pre-processing feedback and applying AI solely for summary or presentation purposes is recommended - with a healthy dose of caution and critical oversight.
AI for Learning: Helping Foundation Year Students in the Physical Sciences use ChatGPT Effectively
AI in Education