University students’ sequential self-regulatory learning actions for completing cognitive-demanding and cognitive-less-demanding tasks with GenAI: ordered network analysis

This study diverges from measuring university students' self-reported perceptions of Generative Artificial Intelligence (GenAI) by investigating their real-time self-regulatory learning (SRL) actions of interacting with GenAI. Self-regulatory learning actions refer to the process by which learners transform their mental abilities into using GenAI to achieve academic goals. Seventeen Chinese university students engaged in two tasks utilizing with the use of ChatGPT: one task involved translating a poem, characterized by a lower cognitive demand, and whereas the other entailed essay writing, which involved a higher cognitive demand. Students' computer screens were recorded, and a think-aloud protocol was implemented to capture both their cognitive processes and behaviours during the tasks. A total of 380 learning actions were recorded to understand how students self-regulated their learning with GenAI, and these actions were subsequently visualized using ordered network analysis. Although Zimmerman proposed three sequential SRL stages (forethought, performance and self-reflection), our findings indicated that university students did not sequentially follow these stages. These findings challenge existing teaching interventions that engage learners in thinking processes across sequential stages of SRL.