<![CDATA[The integration of generative artificial intelligence into higher education has significantly impacted how students engage in learning activities, especially in programming courses. This study investigates how students regulate their AI interaction behavior across different levels of learning difficulty. We collected data from 183 undergraduate students enrolled in the Java Programming course, distributed across six classes. This study analyzes the frequency of prompting, behavioral change during the difficult and easy learning phases, and the primary purpose of chatbot use. The findings show that most students actively adapt their interactions with generative AI based on their perception of the level of learning difficulty. 78 percent of students increase the frequency of prompting when facing challenging tasks, while 69 percent reduce interaction after reaching conceptual understanding. Code generation is identified as the primary purpose of using chatbots, followed by debugging and conceptual clarification. This pattern shows that generative AI serves as a flexible cognitive support tool that students strategically use to manage learning challenges. This research indicates that adaptive prompting behavior represents a new form of self regulated learning in AI supported education environments. Instead of relying on chatbots continuously, students use them selectively to support problem solving and conceptual understanding. This study contributes to the growing body of research on artificial intelligence in education by highlighting behavioral adaptations in students’ interactions with generative AI systems and their implications for programming education in higher education]]>
Copyrights © 2026
