Pragmatic users and skeptical nonusers: A qualitative typology of ChatGPT adoption in physics education

Physics students’ use of ChatGPT or other generative AI tools is a topic broadly discussed in physics faculties around the globe, with discussion often outpacing evidence. What remains unclear is how physics students truly engage with these tools and by what perspectives their choices are shaped. This study develops a data-driven typology of how students perceive the tool’s usefulness and its risks. Through a qualitative content analysis of 1189 survey responses, followed by latent class analysis (LCA), we identified two distinct user profiles. The majority (70%) are “Pragmatic Users” who, while aware of the tool’s inaccuracies, use it for scaffolding tasks like conceptual clarification and finding starting points for problem-set questions. The minority (30%) are “Skeptical Non-Users,” who avoid the tool over concerns about overreliance and its potential to hinder independent problem-solving. These profiles show students making a calculated trade-off between usefulness and risk, suggesting one-size-fits-all policies on AI are inadequate. Thus, rather than promoting uniform adoption, a differentiated pedagogy could create space for students to make informed, self-determined choices—supporting reflective AI use for those who opt in, while equally validating and enabling the learning practices of those who choose not to use generative AI.