CCSE Christmas Seminar 2026

Please arrive early for mingling, coffee, tea, and cookies.

The rapid evolution of generative and agentic artificial intelligence has raised fundamental questions about what it means to be a practicing research scientist. The most striking examples of disruption are in mathematics -- for example, the OpenAI release of a collection of hundreds of AI-generated proofs for problems spanning a huge range of topics in mathematics – but the impact is being felt keenly in physics as well. In this talk I will look at the trends in generative AI, discuss how current hardware trends may impact future evolution of this area, and talk about ways that physics as it is practiced and taught will have to change in order to adapt.

Brian O’Shea is a professor at Michigan State University and a computational astrophysicist and plasma physicist. His research uses numerical simulations to explore galaxy formation and plasma turbulence, and he also studies the teaching and learning of computation in science.

Linear algebra is a key branch of mathematics where central concepts derive meaning from their relationships to one another, and yet students often encounter these ideas through extensive symbolic manipulation and calculations, with limited opportunities to investigate the relationships that connect them. This talk draws on projects that explore how programming can support students in building and connecting those meanings. Within computational environments, students can generate examples, vary parameters, observe patterns, test conjectures, generalize algorithms, and investigate mathematical structures in ways that are difficult to achieve through paper-and-pencil work alone. Coding also provides a source of epistemic feedback; students can act on mathematical objects, examine the consequences of those actions, and use the resulting information to refine their thinking.

Designing for this kind of engagement raises important questions about teaching and learning. Productive work in computational environments often depends on experimentation, iteration, and tinkering, practices that may be unfamiliar to students whose previous mathematical experiences have emphasized obtaining correct answers and utilizing provided algorithms. Drawing on examples from teaching experiments, I discuss instructional structures that support students' exploration while maintaining attention to important mathematical ideas. Throughout the talk, I illustrate how computational tasks can help students coordinate symbolic, geometric, and computational perspectives on core linear algebra concepts. I conclude by considering broader implications for the design of undergraduate mathematics learning environments in which computation functions as the environment for mathematical learning.

Sarah Castle is an assistant professor of mathematics education at the University of Idaho. Her research explores how computation can support mathematical creativity and broaden opportunities for students to engage in mathematics, with a particular focus on using programming to investigate and connect core concepts in linear algebra.

Coffee, tea, and refreshments

Norway’s 2020 curriculum reform introduced programming as a tool for exploring and modelling scientific phenomena. Turning this ambition into sustained change in classroom practice, however, involves more than the technicalities of learning how to code.

This talk presents an overview of my PhD project, which investigates how Norwegian upper-secondary science and physics teachers have interpreted and enacted the curriculum changes regarding programming. Drawing on survey and interview data, I examine teachers’ programming backgrounds, the challenges they encounter, and how they integrate programming into their instruction. I highlight differences between science and physics, exploring how disciplinary traditions, teaching resources, professional development, and formal assessment shape the integration process.

Across the studies, a central theme is the importance of connecting programming to the disciplinary content, practices, and purposes of each subject. I discuss how teachers’ initial efforts have developed over time, what supports or limits continued development, and how generative AI may open new possibilities. The talk concludes with implications for supporting teachers and translating the ambitions set by the curriculum into meaningful classroom practice.

Henning Vinjusveen Myhrehagen is a PhD student at CCSE and KURT (Centre for Teaching and Learning in Science).

Generative AI has quickly become part of how students study. Surveys across countries show that most students now use it, and its use keeps growing. At the same time, AI models can solve many of the problems that physics and other science courses ask students to work through, often better than the students themselves. This raises fundamental questions for teaching and learning: If AI can do the work, what should students still learn, and why? And what should teaching focus on? This talk approaches these questions in four steps: 1) how students use AI, including recent data from Norway, 2) what AI can do in physics, based on a comparison of AI models with Physics Olympiad participants, 3) how students judge AI-generated feedback, compared with how correct it actually is, and 4) what students need in order to use and learn with AI effectively, drawing on research on when people trust what they are told, and what it takes to notice when it is wrong. The talk closes with implications for teaching and learning across disciplines.

Paul Tschisgale is a postdoctoral researcher at the Leibniz Institute for Science and Mathematics Education (IPN). His research explores physics problem-solving and artificial intelligence in education, including the capabilities of large language models and how AI-generated feedback can support students’ learning.