生成式人工智能技术的普及正在深刻重塑汉语自主学习的认知生态。AI工具以“学习搭子”的角色嵌入学习过程，在提升效率的同时，也引发了关于技术依赖与认知能力消解的深层忧虑。本研究主要讨论了AI工具使用对汉语自主学习中学生元认知能力的影响，通过认知卸载理论、体验学习理论与自我调节学习框架，系统考察AI介入如何改变学习者元认知的知识结构、调控过程与发展轨迹。研究发现，AI工具的使用在元认知层面呈现双重效应：一方面，策略性的人机协同可促进元认知知识的外显化与调控行为的可视化；另一方面，过度依赖则可能导致计划能力弱化、监控敏感性下降、评估深度不足及反思性思维的退化。这种“赋能”与“失能”的辩证关系提示，AI对元认知能力的影响并非技术本身所决定，而取决于学习者是否保有认知自主权、是否经历必要的认知摩擦。

The proliferation of generative artificial intelligence technologies is profoundly reshaping the cognitive ecology of self-directed Chinese learning. AI tools, embedded in the learning process as "study companions," enhance efficiency while simultaneously raising deep concerns about technological dependence and the erosion of cognitive abilities. This study examines the impact of AI tool use on the metacognitive abilities of students engaged in self-directed Chinese learning. Drawing upon cognitive offloading theory, experiential learning theory, and the self-regulated learning framework, this paper systematically investigates how AI intervention alters the knowledge structure, regulatory processes, and developmental trajectories of learners' metacognition. The findings reveal a dual effect of AI tool use at the metacognitive level: on the one hand, strategic human-AI collaboration can facilitate the externalization of metacognitive knowledge and the visualization of regulatory behaviors; on the other hand, excessive reliance may lead to the weakening of planning abilities, diminished monitoring sensitivity, insufficient depth of evaluation, and the degradation of reflective thinking. This dialectical relationship between "empowerment" and "disablement" suggests that the impact of AI on metacognitive abilities is not determined by the technology itself, but rather by whether learners preserve cognitive autonomy and whether they experience necessary cognitive friction.