生成式人工智慧的最新發展正在深刻地改變語言學習的樣貌。本研究探討人工智慧，特別是 ChatGPT，在法語為外語（French as a Foreign Language, FLE）書面產出活動中，是否能作為維高斯基（Vygotsky）「近側發展區」（Zone of Proximal Development, ZPD）內的「更有能力他者」（More Knowledgeable Other, MKO）。研究對象涵蓋臺灣所有設有法語系或同等課程之大學（淡江大學、政治大學、輔仁大學、中央大學、文藻外語大學及文化大學），並採用混合研究方法，結合問卷調查、半結構式訪談以及書面產出文本分析。
量化資料（來自學生與教師問卷）透過信度分析、相關分析及跨校比較進行處理，而訪談資料則採主題式編碼分析。研究結果顯示，人工智慧作為一種認知中介，能促進反思性的知識共構與語言自主性，進而擴展學習者的近側發展區。然而，人工智慧並無法取代教師的角色，教師仍然是具有意圖性與社會—情意向度的重要中介者。
本研究提出在混合學習環境中對「更有能力他者」概念的重新界定：ChatGPT 可被視為一種功能性中介（兼具認知與程序層面），而教師則體現為一種批判性與反思性的中介。這些研究發現為批判性數位教學法以及人工智慧在法語為外語教學中的倫理性整合，開啟了新的研究與實踐方向。
Recent advances in generative artificial intelligence are profoundly transforming language learning. This research examines the role of artificial intelligence, and specifically ChatGPT, as a More Knowledgeable Other (MKO) within Vygotsky’s Zone of Proximal Development (ZPD) in the context of written production activities in French as a Foreign Language (FLE). Conducted across all Taiwanese universities offering French as a department or equivalent program (TKU, NCCU, FJU, NCU, Wenzao, and CCU), the study employs a mixed-methods approach combining questionnaires, semi-structured interviews, and analyses of written productions. Quantitative data (from both student and teacher questionnaires) are analyzed using reliability analyses, correlations, and inter-university comparisons, while interviews are subject to thematic coding. The results show that AI functions as a cognitive mediator fostering reflective co-construction and linguistic autonomy, thereby extending learners’ ZPD. However, it does not replace the teacher’s role, who remains an intentional and socio-affective mediator. The study proposes a redefinition of the MKO concept in hybrid learning environments: ChatGPT operates as a functional mediator (cognitive and procedural), whereas the teacher embodies a critical and reflective mediator. These findings open avenues for critical digital pedagogy and the ethical integration of AI in FLE teaching.