Abstract

This talk presents a teaching and assessment approach that incorporates generative artificial intelligence (GenAI) into learning while maintaining a focus on students' developing understanding. Drawing on a portfolio-and-interview model implemented in Mathematical Statistics, the approach combines longitudinal portfolios of learning evidence with individual oral interviews. The portfolio records both material collected from external sources, including GenAI where appropriate, and evidence of students' engagement with that material, while the interview provides a complementary opportunity for students to explain and defend their learning. The model is informed by Fink's taxonomy of significant learning, research on desirable difficulties and generative learning, and the AI Assessment Scale. Rather than relying on AI detection or treating GenAI use itself as evidence of misconduct, the approach emphasises transparent GenAI use and the cognitive work that remains the responsibility of the student. The talk describes the creation and assessment of portfolio content, the role of interviews, and practical considerations relating to implementation.