I am an enthusiastic teacher and interested in didactics of higher education in mathematics. One of my main current interests in this direction is the role of Artificial Intelligence in university education, and specifically mathematics education. The development of the mathematical capabilities of Large Language Models (LLMs) has been advancing with breathtaking speed. This brings, at the same time, big risks and great opportunities for university education in mathematics. On the one hand, students are already using these tools frequently in their everyday work, and since they are capable of solving exercise problems automatically, there is a realistic risk that many students will (or already do) over-rely on these tools and thereby miss their chance to enter into the productive struggle which is so crucial to reach a deeper and intuitive understanding of the subject. In my research I try to think of ways of addressing this important issue, and to come up with and scientifically evaluate suitable adaptations to the classical mathematics lecturing and assessment style. On the other hand, AI also offers a huge capacity to scale personalized tutoring, something which the classic lecturing style in mathematics, simply due to the large student to lecturer-ratio, could previously not offer. Thus, I am also experimenting with and evaluating to what degree suitably designed AI-Tutors (and what types of such tutors) can be effective to engage mathematics students in a critical reflection and interactive discussion of the lecture contents and new mathematical concepts. Below are some links and more concrete projects I am working on.
-Together with Alexander Caspar and Laura Kobel-Keller, we are organizing the seminar Teaching Mathematics in the Era of AI in the HS 26. Our goal with this seminar is to provide a forum for discussing experiences and best practices for using AI in a mathematics lecture, and for this have invited various speakers with significant expertise and prior experience in this wide topic, both from the department as well as other institutions.
-In the summer of 2026, I have been part of evaluating and providing feedback to the first pilots of AI Tutors of the ETH AI Platform, a project that tries to offer flexibly designable and multi-purpose AI-Tutors for all lecturers at ETHZ. I am developing three types of such tutors (prompted to interact with students in different didactical modes) and offer as well as evaluate their usability as part of "conversation-exercises" for my lecture "Structural Graph Theory" in the HS 26. This means that one of the classical written homework exercises is replaced by a conversation task where students are expected to discuss a key new mathematical concept with one of the provided AI Tutors.
-As part of the same lecture, I will also employ "review-exercises". The idea here would be that students are provided with several "proofs" for a given mathematical claim, and are tasked to provide for each of the claimed proofs a review, evaluating the correctness of the solutions as well as where exactly the potential issues lie. This is motivated by the fact that the relevance of the skill to critically evaluate mathematical writing and potential pitfalls has sharply increased with the usage of LLMs by the students.
Other useful links related to the topic:
-MIT Report on AI Use in Teaching, Learning, and Research Training