Artificial intelligence is increasingly embedded in education, yet there remains a need to explain how students conceptualize AI's role in their thinking and learning. This study proposes a framework positioning AI as a dynamic cognitive partner whose function shifts across learning situations. Using qualitative analysis of written responses from 133 secondary students in Hong Kong following completion of an AI literacy course, we identified nine interrelated dimensions through which learners described AI as partnering with their cognition: conceptual scaffolding for difficult ideas; feedback and error detection; idea stimulation; cognitive organization; adaptive tutoring support; metacognitive monitoring support; task and cognitive load regulation; learning continuity beyond classroom boundaries; and explanation reframing through representational flexibility during moments of being stuck or overwhelmed. Across these dimensions, students distinguished between productive support that extends understanding and unproductive reliance that replaces cognitive effort, indicating situational awareness of when AI should and should not be used. Grounded in sociocultural theory, distributed cognition, self-regulated learning, and cognitive load perspectives, the framework clarifies how AI becomes integrated into learners' cognitive activity while illuminating the boundary between cognitive extension and substitution.

翻译：人工智能日益融入教育领域，然而仍需要阐释学生如何概念化人工智能在其思维与学习中的角色。本研究提出一个框架，将人工智能定位为动态认知伙伴，其功能随学习情境而变化。通过对香港133名中学生在完成人工智能素养课程后的书面回答进行质性分析，我们识别出九个相互关联的维度，学习者通过这些维度描述人工智能如何与自身认知协作：为复杂概念提供概念支架；反馈与错误检测；激发思维；认知组织；适应性辅导支持；元认知监控支持；任务与认知负荷调节；超越课堂边界的学习连续性；以及在思维受阻或负荷过重时通过表征灵活性重构解释。在这些维度中，学生区分了能拓展理解的**生产性支持**与替代认知努力的**非生产性依赖**，表明他们对何时应使用或不应使用人工智能具有情境意识。该框架根植于社会文化理论、分布式认知、自我调节学习及认知负荷视角，阐明了人工智能如何融入学习者的认知活动，同时揭示了认知延伸与认知替代之间的界限。