Programming Challenges and Perceptions: A Study of Separate Groups before and after the Release of ChatGPT

Learning to program a new language presents persistent challenges such as debugging, program design and understanding advanced programming concepts, all of which contribute to high dropout rates and learning difficulties. Meanwhile, the emergence of Large Language Models (LLMs) like ChatGPT introduces new dimensions in programming education, offering personalised and interactive learning experiences that could mitigate traditional learning challenges. This study investigates students’ perceptions of programming learning difficulties before and a year after the release of ChatGPT. Through self-reported surveys, the study examines differences in perceived difficulties, evaluates the role of learning materials, including ChatGPT, and explores correlations between students’ perceptions of fairness, trust and effectiveness of ChatGPT and their programming learning difficulties. Findings confirm that even in the era of ChatGPT, the challenges associated with learning programming are complex and continue to be deeply linked to understanding programming concepts, learning situations and learning materials. There was no gender difference in programming learning difficulties, while students concerned about plagiarism from ChatGPT reported more programming learning difficulties. Students who trust ChatGPT report higher perceived effectiveness of its use in learning. These findings highlight the potential of ChatGPT to complement traditional learning resources while emphasising the importance of foundational skills and ethical guidelines. Educators and policymakers must consider these findings to develop programming curricula that leverage technological advancements and address the persistent challenges that hinder student programming learning.