Exploring undergraduates’ computational thinking and human-computer interaction patterns in generative progressive prompt-assisted programming learning

Abstract

Programming education is burgeoning, but it encounters hurdles in implementing thinking-based intelligence instruction. The emergence of generative artificial intelligence, through the utilization of prompt engineering, not only provides meticulous feedback but also significantly elevates the quality and efficiency of human-computer interaction (HCI), thereby nurturing computational thinking (CT). This study aimed to reveal the developmental characteristics of CT and the “black box” of the HCI process through learning analytics methods (i.e., microgenetic analysis, lag sequential analysis, cluster analysis, paired t-tests). 44 college students participated in progressive prompt-assisted programming learning. The results indicated that generative progressive prompts significantly improved students’ CT and its sub-dimensions (i.e., creativity, problem-solving, algorithmic thinking, critical thinking, and cooperativity). Moreover, algorithmic thinking was identified as the core skill in CT development. Additionally, regarding students’ HCI patterns, students with low-level CT focused on more superficial interaction patterns, such as guided and exploratory behaviors, while students with high-level CT concentrated on more in-depth interaction patterns, including debating and summarizing behaviors. Based on our findings, educators in programming should incorporate generative prompts and tailor strategies to accommodate diverse HCI patterns among students with varying CT levels.