Physics Education Research Conference 2026 Plenary Speakers
Plenary Speakers
Parallel Session Types
Submit a Contributed Poster
Submit a Community Connection
Register for PERC
PERC Proceedings Info
Tomasz Bednarz, NVIDIA Corporation
Plenary talk title: Physics ML, Physical AI - Simulating Complex Physics Problems
Tomasz Bednarz is Director of Strategic Researcher Engagement at NVIDIA Corporation, where he works at the intersection of high-performance computing, artificial intelligence, and scientific research. With a background in computational science and interdisciplinary collaboration, he partners with researchers across domains to accelerate discovery and advance the use of advanced modeling and simulation in complex systems.
Bor Gregorcic, Uppsala University
Bor Gregorcic is a researcher at Uppsala University whose work focuses on the intersection of physics education research and emerging AI technologies. He has been a leading voice in examining how large language models can be used and evaluated in physics education, with particular attention to assessment, feedback, and student learning.
Talk title: How AI Became Good at Physics
Abstract: Today, AI chatbots based on large language models and vision-language models can solve many physics tasks remarkably well. This was not always the case. This talk traces the evolution of AI performance on conceptual physics tasks, illustrating the progress through examples drawn from research on the topic. It introduces, in a non-technical way, some of the techniques that have enabled AI systems to become increasingly effective at handling physics problems and offers a reflection on these techniques from the perspective of physics education research.
https://www.uu.se/en/contact-and-organisation/staff?query=N15-929
Odis Johnson, Jr., Johns Hopkins University & ETS
Odis Johnson Jr. is a Bloomberg Distinguished Professor of Social Policy and STEM Equity at Johns Hopkins University, with appointments across education, public health, and sociology, as well as an Edmund W. Gordon Chair of Policy Research and Evaluation at ETS. His work examines the intersections of educational policy, data science, and social inequality, with a focus on advancing equity and justice in education systems. Johnson is also a leading voice on the implications of artificial intelligence for policy, studying how AI systems shape issues such as educational equity, surveillance, and structural inequality, and advancing frameworks for responsible and equitable use of AI in public institutions. He serves as the Edmund W. Gordon Chair for Policy Research and Evaluation at the Educational Testing Service (ETS), where he helps guide research on equitable assessment and educational policy.
Talk title: AI and Educational Policy in an Unprecedented Future
Abstract: How education and schooling from the earliest grades to graduate degree programs will change and function within the age of "intelligentization" or artificial intelligence (AI) is the most important social policy question before our nation. While physics education is perhaps ahead of other education subfields in developing and integrating AI-based instructional technologies, this address asks us all to consider the educational systems in which those AI innovations will be implemented - the transformation they must undertake with arguably limited guidance and while facing historic challenges, both institutional and societal. Within this context, this address identifies the critical questions about AI systems integration that must be answered for education policy to prepare us for the unprecedented future.
https://education.jhu.edu/directory/odis-johnson-jr-phd/
Tor Ole B. Odden, University of Oslo
Tor Ole Bigton Odden is a researcher at the University of Oslo whose work focuses on advancing methods in physics education research through the use of artificial intelligence. He develops and applies natural language processing techniques to enable large-scale qualitative analysis, including the automated coding of student responses and the study of patterns in PER texts.
Talk title: Making Sense of Chat-Based LLMs in Physics Education: Role-Playing Machines, Blurry JPEGs, and Conceptual Blenders
Abstract: What is ChatGPT actually doing when it explains physics? Why can it produce a helpful analogy one moment and an authoritative mistake the next? And how can we, as physics education researchers, make sense of these systems well enough to use them productively?
PER is unusually well set up to answer these questions. We have a long history of using theoretical frameworks and toy models to understand complicated knowledge systems that we do not have direct access to. Chat-based LLMs are, in many ways, another such system: strange, powerful, opaque, and constantly surprising. In this talk, I will introduce three theoretical frameworks that I find useful for making sense of them: the Role-Playing Machine, the Blurry JPEG, and the Conceptual Blender. Using these frameworks, I will unpack some of LLMs' recurring strengths and weaknesses, like why they are effective at qualitative explanation, analogy, and reformulation but unreliable with precise facts, specialized knowledge, and mathematical nuance. My goal is to show how PER can use its intellectual tools to build intuition about when LLMs may support physics learning, when they may undermine it, and how we might study their effects on students' engagement with physics.
