Article Preview
TopIntroduction
The rapid advancement of generative artificial intelligence (AI) represents a major paradigm shift in educational technology (Ouyang & Jiao, 2021). Unlike traditional systems that functioned as passive tools, contemporary AI demonstrates advanced generative and adaptive capabilities, enabling it to interpret task intentions, analyze linguistic structures, and produce semantically coherent, human-like text (Kasneci et al., 2023). This progression has shifted AI from a “technological intermediary” to an intelligent participant in teaching, fostering a new learning mechanism defined by human–machine collaboration (Ouyang et al., 2022). In higher education, these trends are evident in academic writing, where generative AI increasingly acts as a collaborative agent (Delgado-Kloos et al., 2024, 2025). AI writing tools support students not only with grammar correction and formatting but also with higher-order processes such as idea generation, structural organization, and logical reasoning (Marzuki et al., 2023; Yan, 2023). As a result, writing has shifted from a student-driven independent activity to a human–machine collaborative process with deep technological integration. Within this transformation, writers’ psychological states and agentic behaviors are being reshaped, with their autonomy, strategy regulation, and motivational orientation undergoing unprecedented changes and challenges.
Although AI tools can enhance writing efficiency and confidence, research also reveals negative effects of AI dependency. Gerlich (2025) found that frequent AI users exhibited a significant negative correlation between critical thinking ability and reliance on AI, tending to delegate cognitive responsibility during complex reasoning and text-construction tasks, thereby weakening analytical and reflective capacity. While delegating core tasks to AI may facilitate task completion, it reduces cognitive engagement and hinders skill development (Gerlich, 2025; Risko & Gilbert, 2016). As cognitive tasks are increasingly outsourced to technology, learning risks shifting from capability development rooted in cognitive training to tool operation focused on outcomes. This shift has tangible consequences, particularly declining performance in non-AI contexts and implications for educational equity. For example, students may score higher in AI-assisted tasks but perform worse without support, undermining long-term confidence and engagement. Marzuki et al. (2023) observed that excessive reliance on AI writing tools constrains critical thinking and originality. Similarly, Vieriu and Petrea (2025) found that AI use diminishes autonomy and creativity, impairing academic development. Overreliance may also exacerbate inequality: lower-ability students may gain short-term benefits from AI, but these gains fail to cultivate independent thinking or writing competence, thereby widening disparities and deepening stratification in educational resources.