人工智慧應用於自主學習活動設計

Applying AI to the Design of Self-Regulation Learning Activities

在人工智慧（Artificial Intelligence, AI）快速普及與數位學習生活化的趨勢下，學習成效的關鍵不再僅是資源取得，而在於學生能否形成可遷移的自主學習能力。然而，在AI高度可得的學習情境中，如何將自主學習策略轉化為可教、可練、可檢核的活動設計，並釐清AI之支持角色與介入界線，仍待探究。本文以Zimmerman自我調節學習理論為基礎，整理10項可操作之自主學習策略，並將AI定位為可選擇、可撤除之學習鷹架，透過提問、提示、檢核與回饋等支持策略運用與後設認知監控，促進自我評估校準。另以國小六年級速率單元為例，嵌入「定標—擇策—監評—調節」歷程，呈現AI支持下自主學習策略介入時機與角色。最後提出兩項教學設計建議：重點在教會學生自主學習策略；AI採「低介入、高精準」，並配合撤除與責任移轉，以避免依賴與能力錯覺。

With the rapid diffusion of artificial intelligence (AI) and the normalization of digital learning, learning effectiveness increasingly depends on students’ transferable self-directed learning capabilities rather than mere access to resources. However, in AI-rich environments, it remains unclear how self-directed learning strategies can be operationalized into teachable, practicable, and assessable activity designs, and how the roles and boundaries of AI support should be delineated. Grounded in Zimmerman’s self-regulated learning (SRL) framework, this paper synthesizes ten actionable self-directed learning strategies and conceptualizes AI as an optional and fadeable scaffold. Specifically, AI supports learners’ strategic engagement and metacognitive monitoring through questioning, prompting, checking, and feedback, thereby enhancing the calibration of self-evaluation. An instructional example from a sixth-grade mathematics unit on rate is provided to illustrate the integration of an SRL cycle– goal setting, strategy selection, monitoring and evaluation, and regulation– and to clarify the timing and functions of AI-mediated support. Two instructional implications are proposed: prioritizing strategy instruction over tool adoption and adopting low-intervention, high-precision AI scaffolding with scaffold fading and responsibility transfer.