This study aims to investigate the relationships between dissertation self-efficacy, academic buoyancy, flow experience, and persistence intention among doctoral students, and to examine the moderating role of AI usage in AI-enhanced academic writing.
Data were collected from 426 doctoral students using purposive sampling. Exploratory factor analysis was conducted to identify the dimensions of dissertation self-efficacy, and structural equation modeling was used to test the hypothesized relationships and the moderating effects of AI usage.
The study found that AI usage had a significant negative moderating effect on the relationship between flow experience and persistence intention. Additionally, dissertation self-efficacy, consisting of research implementation, self-management, and research competence, significantly influenced academic buoyancy and flow experience, which in turn affected persistence intention.
The cross-sectional design of the study limits the ability to draw causal conclusions. The sampling method may introduce survivorship bias, and the measurement of AI usage does not differentiate between specific applications or usage patterns. Future research should track students over time and capture detailed metrics on their AI usage to better understand its impact.
AI usage impedes doctoral persistence. Students who heavily use AI are less likely to complete their degrees. This finding requires immediate institutional action. Universities should teach doctoral students that while AI can assist with routine tasks, overreliance undermines critical thinking and problem-solving essential for doctoral success. AI should support, not substitute, doctoral education.
This study extends Social Cognitive Career Theory by integrating academic buoyancy and flow theories in the doctoral context and provides new insights into how AI usage can undermine the psychological mechanisms that drive doctoral students’ persistence.
