Exploring the Computational Thinking Process of College Students: Collaborative Programming with LLMs

Computational thinking (CT) has gained increasing attention from educators and practitioners due to its effectiveness in fostering student development, particularly in programming courses. Meanwhile, with the rapid advancement of large language models (LLMs), an increasing number of teachers and students are incorporating AI-assisted teaching and learning. However, empirical research on the patterns of CT processes when students adopt different AI-assisted learning strategies remains limited. This study explores the differences in CT process patterns when students use two different types of LLMs: General AI Group (GAIG) and No Code AI Group (NCAIG). Results showed that: (1) The social perspective is the most dominant in CT (2) Questioning and connecting co-occur most often, followed by connecting and sequencing, as well as questioning and sequencing; (3) Students' prior knowledge significantly affects their CT patterns. These findings provide empirical support for educators to effectively guide students in leveraging AI technology to enhance CT in instructional settings.