Scalable Classification of Course Information Sheets Using Large Language Models: A Reusable Institutional Method for Academic Quality Assurance
Researchers developed an end-to-end pipeline using large language models to scan thousands of university course information sheets. The method classifies courses into risk tiers (Clear, Potential, Low) regarding student use of GenAI tools. Iterative prompt engineering refined the model for operational use, showing that risk classifications improved over time and were most accurate for practical programs.