A study of the effect of peer-led instruction on the improvement of swimming skills of university students.

Generative AI, exemplified by ChatGPT, possesses capabilities in massive data processing, exceptional language understanding, and text-to-image generation. By enabling intelligent writing, optimizing academic searches, and assisting with routine tasks, it has emerged as an important research aid for graduate students (). As artificial intelligence becomes increasingly embedded in academic research, the phenomenon of graduate students’ excessive dependence on GenAI has attracted growing scholarly concern. In this study, GenAI dependence is not treated as a clinical form of addiction, but as a continuum of reliance on generative AI tools in academic and research-related tasks. This continuum may include both functional reliance, such as using GenAI to improve efficiency and support research performance, and more problematic dependence, such as relying on GenAI to reduce anxiety, avoid cognitive effort, or substitute for independent academic judgment. Therefore, a higher level of GenAI dependence does not necessarily indicate maladaptive use in all cases; its meaning depends on the psychological conditions under which it emerges, such as research anxiety, performance expectancy, and critical thinking. Existing research suggests that GenAI use may have both enabling and risk-related implications. On the one hand, GenAI reflects the instrumental rationality of technological empowerment and efficiency-seeking. On the other hand, excessive or unreflective reliance on GenAI may be associated with academic inertia and may potentially weaken graduate students’ academic autonomy, innovative thinking, and problem-solving abilities during their formative academic years (). Prior studies have also noted that the growing use of ChatGPT and other GenAI tools may be accompanied by dependence-related risks, including misuse and overreliance (–). Under high-pressure and competitive research environments, a sense of time urgency has become a prevalent psychological state among graduate students (). When perceived time pressure coincides with the accessibility of GenAI tools, students may regard GenAI as a convenient means of accessing information, completing academic tasks, and responding to pressing research demands (). In this context, time pressure may be positively associated with students’ reliance on GenAI in their academic work (). However, this association may operate through different psychological pathways. Some students may experience research anxiety under time pressure and turn to GenAI for emotional relief, whereas others may view GenAI primarily as an efficiency-enhancing tool for achieving performance goals. The extent to which different types of GenAI-dependent users show higher levels of research creativity may depend on their higher-order cognitive capacities, such as critical thinking (). To capture individual differences, this study employed a mixed-analysis strategy, integrating PROCESS-based mediation analysis to examine the indirect associations linking time pressure, research anxiety, performance expectancy, and GenAI dependence, and latent profile analysis (LPA) to identify the GenAI dependence profiles of graduate students. Furthermore, it analyzed the demographic and cognitive correlates of these profiles and the differences in research creativity across different profiles. This study systematically addresses the following questions: First, how is time pressure associated with graduate students’ GenAI dependence through research anxiety and performance expectancy? Second, what GenAI dependence profiles can be identified among graduate students, and what characteristics define each profile? Third, which demographic and cognitive factors are associated with profile membership? Fourth, how does research creativity differ across different GenAI dependence profiles? Time pressure refers to graduate students’ perception of time scarcity in completing academic and research-related tasks (). Under the framework of Social Acceleration Theory, higher education has increasingly been characterized as “accelerated academia” (). The economic logic of efficiency-seeking, the structural compression of academic time, and the cultural engine of time aversion collectively propel graduate students’ academic activities into a state of acceleration, thus turning to technological dependence to regain a sense of temporal efficiency (). Within this context, graduate students may adopt different stress-coping strategies, including emotion-focused coping and problem-focused coping (). When graduate students experience time pressure, on one hand, they develop research anxiety, defined as the negative emotions such as tension, unease, and irritability arising from perceived specific or potential threats related to unfinished research tasks, and may turn to on GenAI to alleviate the discomfort associated with anxiety (). Other students may perceive GenAI primarily as an efficiency-enhancing tool and may use it to support academi