AI for Learning: Helping Foundation Year Students in the Physical Sciences use ChatGPT Effectively

The ‘AI for Learning’ project set out to explore how students on the Astrophoria Physical Sciences Foundation Year programme could engage with ChatGPT in ways that enhance, rather than undermine, their academic development. With the ultimate aim of supporting constructive and active learning, the project focused on training students on how to use ChatGPT Edu, and training tutors to use it so they could assess its accuracy in their subject area and develop a student guide to help learners use AI as a complement to their own skills.

Supported by Kelly Webb-Davies, Lead Business Technologist in the AI Competency Centre, tutors explored three methods of teaching to create custom GPTs for. These included:

1. A Socratic tutor that encouraged students to think through problems rather than simply receive answers, mirroring Oxford’s tutorial style;
2. A self-quizzing tool to identify gaps in knowledge and deepen conceptual understanding;
3. A paragraph challenge, in which students had to spot hidden errors in AI-generated text, promoting critical reading and reflection.

Student responses to AI were, at first, polarised - ranging from scepticism to over-reliance. However, by the project’s end, all participants had developed a more balanced view. Cautious students gained confidence in using AI as a learning partner, while initially over-reliant students better recognised the risks of bypassing their own thinking processes.

Key lessons emerged: successful integration of AI into undergraduate education requires clear guidance to support active learning, and open dialogue about AI use is far more effective than ignoring it. Importantly, students must understand that AI is not a shortcut to understanding - they still need to demonstrate their own abilities in assessments.