Using generative AI to support students’ interpretation of kinematics graphs: a classroom-based approach

Abstract

This paper reports a classroom intervention in which secondary vocational students used generative artificial intelligence (AI) to discuss and check their reasoning with kinematics graphs. The intervention was conducted within the AI4Physics project and consisted of five 90 min workshops. The activities focused on common difficulties such as confusing graph shape with motion, graph height with slope, and the area under a graph with a visual feature rather than a physical quantity. Students worked with position–time, velocity–time, and acceleration–time graphs and were guided to use AI for explanation, comparison and checking rather than for obtaining final answers. Students completed complementary test of understanding graphs in kinematics-based pre- and post-test item sets. Descriptively, their mean score was 2.91/13 on the pre-test item set and 5.50/13 on the post-test item set. Because the two item sets were not independently validated as equivalent and the study involved only one class, the results are interpreted with caution. The distinctive pedagogical contribution of the paper is a teacher-guided checking routine in which AI-generated explanations, including incorrect ones, become objects for physics reasoning: students are required to test AI claims against axes, units, slopes, areas and graph intervals before accepting or rejecting them.