ChatGPT’s problem-solving abilities in context-rich and traditional physics problems

Abstract This study explores the ability of ChatGPT in solving first-level physics textbook-style problems and context rich problems (CRPs). The findings reveal that while ChatGPT demonstrates proficiency in identifying relevant physics concepts and solving traditional problems, it faces challenges in handling CRPs which require accurate modelling and reasonable assumptions. These insights highlight both the potential and limitations of using ChatGPT in Science, Technology, Engineering, and Mathematics education, providing a foundation for future research into leveraging AI to support student learning and human–AI collaboration in addressing real-world challenges.