From Passive Prompting to Critical Partnership: AI Literacy, Human-AI Collaboration, and AI Dependency among University Students

Abstract

As generative artificial intelligence (AI) tools become embedded in everyday academic life, universities face a pressing question: does deeper AI literacy help students collaborate with AI responsibly, or does convenient access to AI simply deepen dependency regardless of what students know about the technology? This narrative review synthesizes recent theoretical and empirical literature, comprising more than 25 peer-reviewed and preprint sources published largely between 2020 and 2026, to examine the relationship between AI literacy, the quality of human-AI collaboration, and AI dependency among university students. The review finds consistent evidence that AI literacy, conceptualized across dominant frameworks as the capacity to understand, critically evaluate, and effectively collaborate with AI systems, is associated with more active, dialogic, and metacognitively engaged patterns of AI use, whereas low AI literacy is associated with passive, answer-seeking interaction patterns that resemble instruct-and-accept transactions rather than genuine collaboration. At the same time, the review identifies an important boundary condition: several recent experimental and psychological studies show that AI literacy and prompting skill alone do not fully immunize students against dependency, cognitive offloading, or sycophantic reinforcement of their own errors, particularly under time pressure or cognitive fatigue. The discussion argues that AI literacy functions as a necessary but not sufficient condition for responsible human-AI collaboration, and that its protective effects are strongest when combined with structured pedagogical scaffolding, self-regulated learning skills, and system-level design choices that resist over-trust. The review concludes with implications for curriculum design, faculty development, and institutional policy, recommending a multidimensional approach to AI literacy education that integrates technical understanding, critical evaluation, ethical reasoning, and explicit self-regulation strategies to help students build genuine collaborative competence rather than convenient reliance.