The cellular correlates and adolescent reorganisation of cortical myelination networks in the common marmoset.

Generative artificial intelligence (GenAI) is rapidly transforming the learning practices of university students. Represented by tools such as ChatGPT, GenAI applications—owing to their powerful capabilities in content generation and human–computer interaction—have become important auxiliary resources supporting students’ self-directed learning (SDL). Recent studies indicate that more than 70% of university students have used GenAI tools to assist with academic tasks, with approximately 40% reporting at least weekly use. By providing extensive learning resources, personalized learning recommendations, and efficient problem-solving support, GenAI tools can enhance learning efficiency and broaden students’ academic perspectives. However, the integration of GenAI into learning processes does not always produce positive outcomes. In practical learning contexts, students’ interactions with AI systems sometimes involve irrational or unconscious usage patterns. One emerging concern is the thoughtless use of generative artificial intelligence (TUGA), which refers to situations in which users adopt AI-generated answers, texts, or ideas without critically evaluating, verifying, or deeply understanding the outputs. Some students may become overly dependent on AI tools and unreflectively copy generated content into academic work. Such practices may hinder the development of critical thinking and problem-solving skills and may also raise concerns about academic integrity. Although GenAI technologies provide significant learning support, excessive or unreflective reliance on these tools may weaken students’ cognitive engagement in the learning process. Several related constructs have been proposed to describe problematic patterns of AI use. For example, Fan et al. introduced the concept of unreflective reliance on ChatGPT, which highlights learners’ tendency to accept AI-generated responses without sufficient metacognitive monitoring. Similarly, Sardi et al., discussed excessive dependence on generative AI, emphasizing behavioral overreliance on AI tools during learning tasks. Other studies have examined the general use of generative AI in educational contexts, which mainly captures the extent or frequency of AI usage rather than problematic usage patterns. In contrast, the concept of TUGA proposed by Hou et al., emphasizes a low-reflection cognitive processing pattern in which users automatically adopt AI-generated outputs without critical evaluation or verification. Compared with general AI usage constructs, TUGA more directly captures a maladaptive interaction pattern between learners and AI systems and is therefore particularly suitable for examining potential negative consequences for learning outcomes such as SDL. Although Hou et al., reported that the proportion of TUGA ranges from approximately 10% to 16.7%, this behavioral pattern may still have disproportionately negative consequences for students’ learning processes.