Session 1 - Strand 1

Abstract:

This is a case study of a time-pressured module revival in which all teaching resources - from handbook to assignment briefs- were generated using AI tools, principally NotebookLM, Gemini, ChatGPT and Claude. The workflow centred on drafting a content outline through Storm or Perplexity, expanding it with a large language model, uploading it to NotebookLM, and exporting student-facing materials to Moodle. Students could interact directly with the sources via NotebookLM’s chat function. Assessment briefs were AI-assisted, and an experimental AI marking workflow was piloted. Friction arose from slide generation complexity, unstable workflows, and institutional IT restrictions that limited student access to some tools. On the positive side, careful source curation eliminated hallucination, markdown proved a reliable common language across tools, and the combination of multiple frontier models outperformed reliance on any single one. The case study closes with reflections on time cost, tool selection, and lessons for future iterations.