Coding Error-Based Reflective Learning with ChatGPT: A Study on Higher-order Thinking Skills in Programming Education

ChatGPT shows potential in programming education, but there are concerns about students becoming overly reliant on it, which could hinder the development of higher-order thinking skills (HOTS). This study introduces a new method that uses ChatGPT to analyze students' coding errors and produce personalized reflective reports. The goal is to enhance HOTS while maintaining independent thinking. Prior research has not adequately explored the use of AI for error-based reflective learning in programming education. The quasi-experimental study investigates the impact of ChatGPT-generated, coding error-based reflective reports on improving critical thinking and problem-solving skills in a Python course. A total of 49 undergraduates participated over two semesters. The experimental group (n=25) received ChatGPT-generated reflective reports based on their coding errors, while the control group (n=24) used traditional reflective methods. ANCOVA analyses revealed significant improvements in the experimental group for both critical thinking (F=6.85, p<0.05) and problem-solving skills (F=8.34, p<0.01). These results suggest that AI-driven, coding error-focused reflective feedback can enhance HOTS in programming education. This study contributes to the field of AI-assisted learning and offers insights into incorporating AI-supported reflective practices in programming courses.