The Impact of GenAI Assistance on Knowledge Building in Tasks of Different Difficulty Levels

Online learning has emerged as the preferred learning modality for numerous learners. However, during this process, learners' discussions are frequently confined to simplistic question-and-answer sessions, thereby impeding the attainment of high-level knowledge building. Generative Artificial Intelligence (GenAI), equipped with its capabilities of personalized content recommendation and knowledge generation, presents novel practical opportunities for facilitating the construction of high-level knowledge among learners. In this study, a sample of 143 in-service graduate students was selected as participants, and epistemic network analysis was employed to investigate the impact of GenAl on learners' knowledge building processes across different levels of learning tasks. The results of the study indicate that GenAI exerts a relatively weak influence on both low-level and middle-level tasks. Under the GenAl-assisted learning model, learners demonstrated more prominent performance in high-level tasks, showcasing stronger individual analytical abilities and autonomous cognitive characteristics. Nevertheless, GenAI might potentially undermine social interaction and collaboration within their learning process. Based on these findings, this study offers empirical evidence and corresponding recommendations for the application of GenAI in classroom instruction.