Pedagogical Impacts of ChatGPT and Other LLMs

Dr Mark Zarb, Robert Gordon University

As we approach the anniversary of our workshop on the “Pedagogical Implications of ChatGPT and other LLMs,” hosted by RGU’s School of Computing and sponsored by SICSA Education, I wanted to take the opportunity to reflect on a day of exciting discussions and perspectives. The event drew a full house of 40 attendees, and included speakers from the RGU Applied Computing Education (ACE) research group, Edinburgh University, the University of Strathclyde, and Falmouth University.

The talks delved into various aspects of integrating Language Models (LLMs) into higher education. Speakers explored the application of these tools in data science education, highlighting how AI can enrich learning experiences with personalized contextual examples. Discussions also tackled the nuanced challenges of AI explainability, emphasizing the importance of transparency in AI-driven decision-making processes. One of the most captivating discussions centred on the propensity for pareidolia in human-AI interactions, examining how humans interpret patterns in AI-generated content. This theme resonated strongly across disciplines, prompting educators from diverse fields to explore the ethical and educational implications of these interactions.

Throughout the day, the spotlight was firmly on ChatGPT and its role in education. What emerged was an exciting discussion about the potentials – and perils – of using these technologies, and the importance of emphasising to students that ChatGPT isn’t an oracle – but a tool. We also spoke about the need to embed topics like ethical and responsible use of ChatGPT throughout the curriculum – a topic that persists in its importance.

I can now see that this workshop not only deepened our understanding of LLMs but also fostered a collaborative environment where interdisciplinary perspectives enriched the dialogue and helped us start to prepare for the unique challenges that generative AI brought to this last academic year. It has also led to significant outcomes, including ongoing partnerships with researchers from Strathclyde, culminating in a workshop at ACM CEP 2024 and numerous submissions to SIGCSE Virtual 2024.

On behalf of the ACE research group at RGU, I wanted to extend our thanks to SICSA Education and all participants for their invaluable contributions to this event.

ACE are always looking for collaborators! For more information on their current work and interests, feel free to talk with Mark at m.zarb@rgu.ac.uk