ChatGPT and DeepSeek in the Physics Classroom: Affordances, Limitations, and Practical Applications

The rapid emergence of large language models (LLMs) is reshaping the landscape of physics education by offering new avenues for conceptual exploration and individualized support. This study examines the classroom affordances and limitations of two prominent LLMs—ChatGPT and DeepSeek—through a comparative analysis of their dialogic, representational, and reasoning capacities. Drawing on recent research in physics and science education, the article investigates how these tools can facilitate inquiry, scaffold problem solving, and enhance inclusive practices while also posing challenges related to bias, accuracy, and teacher authority. Particular attention is given to their complementary strengths: ChatGPT’s conversational flexibility and accessibility, and DeepSeek’s structured reasoning and technical precision. Pedagogical and ethical implications are discussed alongside recommendations for hybrid applications and teacher-led orchestration. The findings provide evidence-informed guidance for educators, curriculum designers, and policymakers seeking to integrate artificial intelligence into physics instruction without compromising critical thinking or equitable access.