At PHYSCON 2026, there was a session of small group discussion on AI in Teaching Physics exploring the issues and opportunities. Each of the groups was asked to provide a record of the discussion, either as an audio file, a transcript of the discussion or as a written report. The audio files were converted to text files and along with the others summarised with AI into the following documents:
- Summary of round table discussions
- SWOT analysis with focus on practical examples of AI use
- An image summarising the SWOT analysis see right.
These articles and videos explore some of the issues on using AI in Teaching Physics.
- This presentation by Derek Muller at the Perimeter Institute in April 2025 on ‘What Everyone Gets Wrong About AI and Learning’ is an important contribution. Derek explores how AI might change how we teach and learn, drawing on insights from past tech shifts and core principles of cognitive science. While AI presents exciting opportunities, it also introduces real risks—especially when it comes to how our brains build knowledge and expertise.
- This interview with Mustafa Suleyman, the co-founder of Deep Mind and CEO of Microsoft AI, with the editor of The Economist on Intelligence Squared is an insightful and eloquent description of benefits and risks of future development of AI and the need to ensure public oversight of that development.
- This article on ‘ChatGPT in physics education: A pilot study on easy-to-implement activities, looks promising.
- This article in Physics Education titled ‘ChatGPT and the Frustrated Socrates’ is a useful exploration of the benefits and potential problems in using ChatGPT.
- AI in Physics Education: a comprehensive literature review. An article in Physics Education.
- Youtube video “Empowering Physics Education with AI” by the AI Science Guy. An 18 min presentation on how AI can be useful in teaching physics, a set of detailed powerpoint slides effectively presented. He also gave a 2 hour lecture, which was recorded from the side, so some slides could not be read, but the audio is fine.
- Creating a customisable Socratic AI physics tutor An article in Physics Education by Tufino and Gregorcic. The article designs a Socratic physics problem-solving tutor using Google’s Gemini Gems feature.
- Teaching with Questions, Not Answers: ChatGPT as a Socratic Partner in Physics Learning An article in Physics Education by Suarez. Conceptual conflict tasks are particularly effective in encouraging critical thinking, as they present mutually incompatible but plausible responses based on common alternative conceptions held by students. These tasks elicit cognitive conflict and uncertainty, which are favorable conditions for productive Socratic dialogue. The presence of ChatGPT as a conversational partner adds complexity: its responses—sometimes correct, sometimes ambiguous or incorrect—prompt students to interrogate claims and identify inconsistencies.
- Using ChatGPT for teaching physics An article in Physics Education by Karina Avila et al. The emergence of chatbots such as ChatGPT presents a unique opportunity for physics teachers to enhance their instructional methods. These advancements allow teachers to provide more effective and personalised educational experiences to their students. They can be used to generate content and interact with students, providing formative feedback that supports various aspects of self-regulated learning. However, the integration into the classroom is not straightforward, as the output is not always reliable and the prompt needs to be well specified.
- Something Big Is Happening An article by Matt Shumer, published in early Feb, 2026, has 80 million views within a month. He is a professional in the field as an app developer. He suggests strategies on how to manage the coming changes in AI. The article has its own wikipedia page, which includes links to some critiques.
- Don’t Call It ‘Intelligence’ An article in The Atlantic by Charles Yu that argues that humans are ‘questioning machines’ while AI is a ‘answering machine’ and what that means for a definition of intelligence. This article is behind a pay wall, but can be accessed by a free trial.
For examples of how physics teachers are using AI:
UK Physics Teachers have a forum ‘Physics Teachers News and Comment’ (PTNC) to which teachers regularly share ideas. VicPhys News 1/T2/24 has more details about PTNC. In 2024 there was a request posted about AI in physics, which received the following replies:
- Generating really lovely cover images for new topics… a summary of Quantum Physics in the style of Rembrandt? perfect. Isaac Newton battling a velociraptor? Just the image I needed. Frivolous but actually very pretty and quite a fun start to a topic.
- Students to try out as a revision tool: “Copilot, please generate 5 multiple choice questions for me to test my understanding of XXX, don’t give me the answers just yet”.
- To help students explore topics- “Please write a story in the style of a fantasy novel to illustrate the ideas of XX”.
- A teacher’s documentation of how they are using AI to generate questions and answers, creative resources and plan lessons (20 pages of the outcomes, but most pages have short entries)
- Use Perplexity AI which gives sources to the student questions such as ‘What questions should I ask myself when ….. ‘. Was really nice for cutting the student ‘research’ time down, and very effective (and the tool is free).
- I included a complicated paper into a task recently and then suggested to students they might have a way to simplify it. Others were already pasting into Chat GPT. However, after a few minutes to make sure everyone had figured it out, I stopped the class and asked people to give their prompts. We then generated a ‘best prompt’ that extracted a simplified summary of the paper with the key points. The students found this really useful both in terms of learning about the AI tool, but also understanding they had another way of accessing difficult material.
- Some teachers are using Gemini, part of Google Apps.