My digital mentor: a mixed-methods study of user-GAI interactions

Generative Artificial Intelligence (GAI) has emerged as a powerful tool in online learning, offering dynamic, high-quality, and user-friendly content. While previous studies have primarily focused on GAI's short-term impacts, such as users' acceptance and initial adoption, a notable gap exists in understanding long-term usage (i.e., infusion use) and the psychological mechanisms.

This study employs a two-stage mixed-methods approach to investigate users' infusion use of GAI in online learning scenarios. A semi-structured interview (N = 26) was conducted in the first stage to develop a systematic framework of influencing factors. These factors include intelligence, explainability, response time, integrability, accuracy, source credibility, personalization, and emotional support. The second stage empirically validated the research framework using survey data of 327 participants. We find that the eight factors influence users' infusion use through two key psychological mediators: perceived value and satisfaction. We also used the fsQCA method to obtain the configurations. These configurations demonstrate that no single factor alone is sufficient; rather, it is the combination of multiple factors that fosters users' infusion use.

Our findings contribute to expanding the literature on the application of the theoretical literature on technology adoption in online learning contexts and provide practical implications for developing effective user-GAI interaction.