The integration of artificial intelligence (AI) in physics education has shown remarkable potential, particularly in learning analytics and personalized feedback for interpreting graphical problems in kinematics and dynamics. Although models such as ChatGPT-4 show promising capabilities in problem-solving, they still face significant challenges in accurately interpreting complex graphical representations. While these tools may support learning and provide immediate feedback, their limitations be