Seminar in Teaching and Learning in Mathematics

The seminar is joint between the two research groups TLM at Mathematical Sciences and PER, Physics Education Research at Physics and Astronomy.

13:00-13:45 Informal presentation with focus on didactics

13:45-14:00 Fika

14:00-14:45 Seminar

Abstract: Recently, it was announced that an internal OpenAI model had solved Erdös’ "unit distance problem" by providing a counter-example. Whether one thinks of such developments as good or bad, it clearly illustrates the growing mathematical capabilities of AI - even at the research level. In this talk, I discuss how generative AI can be used as a force for good in higher mathematics education. Drawing on two years of experience integrating generative AI into first-year courses in Calculus, and informed by ongoing educational research, I argue that a beneficial way forward can be found if we re-calibrate three tensions that are at the heart of education: (1) human versus machine competence, (2) mechanisms for trust versus accountability and (3) individual versus collective learning processes. My central claim is that generative AI introduces no qualitatively new problems in education. Instead, it is making some long-standing problems (that we could previously tolerate) impossible to ignore, while simultaneously offering new ways to address challenges that previously seemed beyond reach.