When Is It Time to Jump? The Boiling Frog Problem of AI Use in Physics Education

From the Journal: The Physics Teacher

WASHINGTON, Jan. 6, 2026 — Generative AI is becoming ubiquitous in everyday life. Large language models like ChatGPT can help answer questions, write emails, and solve problems at seemingly lightning speed, pulling from enormous datasets to engage in conversations with their users. Generative AI tools are increasingly used in classrooms, too, sometimes to supplement learning and sometimes to cut corners.

Because of this, many teachers have expressed concerns about this powerful tool’s impact on student learning and development. In The Physics Teacher, a journal co-published by AIP Publishing and the American Association of Physics Teachers, a physics professor-turned-AI-researcher from ETH Zurich explores the uses of generative AI to teach physical science, both its helpfulness and its hindrance on learning.

In his paper, author Gerd Kortemeyer compares the constantly increasing physics capabilities of generative AI to the boiling frog fable, which predicts that a frog will fail to recognize the danger of a gradually heating pot until it’s too late to hop out.

“Generative AI can absolutely be a helper: for example, as a tool to quickly pull up definitions, explaining terms, drafting analysis programs, giving students immediate feedback on their explanations, or for translating physics concepts into different languages,” Kortemeyer said. “But these are supports for human sense-making and collaboration, not the main act.”

Surveys are showing that students are using AI tools more frequently. Kortemeyer lays out situations where generative AI usage may be warranted and places where it may not help education — and therefore, a “jump out of the pot” is warranted.

Kortemeyer explains that unsupervised, online assignments are no longer viable as examples of mastery, since AI tools can correctly solve homework problems from a picture and AI-detection tools are questionable at best. However, removing AI tools from education in general risks alienating students.

“Change is hard. We have spent years of our lives perfecting our lecture notes and PowerPoint slides, and we have built up beautiful problem banks with neat little tricky scenarios,” said Kortemeyer. “I think as physics educators, we need to completely recalibrate — what do we really want to teach?”

Kortemeyer focuses on ways AI can be integrated into the learning process, so students learn to cite and critique AI usage. These research-based instructional methods will take stamina, but by adapting rather than ignoring, educators can ensure that the water in the pot doesn’t get too hot for the proverbial frog.

“We have brought generative AI on ourselves — physicists played a large role in developing the underlying principles. Now we need to live with the consequences, but that is not necessarily a bad thing,” said Kortemeyer. “If we take it as an opportunity to focus on reasoning, collaboration, and genuine understanding rather than on speed and routine problem solving, we may finally be doing the right thing by our students.”

###

Article Title

The Boiling-Frog Problem of Physics Education

Authors

Gerd Kortemeyer

Author Affiliations

ETH Zurich