從知識共享到協同共創──交融記憶視角下的設計者與生成式AI協作

From Knowledge Sharing to Co‑Creation: Designers’Collaboration with Generative AI from the Perspective of Transactive Memory Systems

本研究從交融記憶系統與知識維度的視角，探討設計學生與生成式人工智慧（generative artificial intelligence, GAI）之間的協作模式及涉及的知識類型。本研究採用日誌記錄法，招募三位設計學生，紀錄其與AI在共同完成設計任務時的互動歷程，分析當中呈現的任務分工、資訊交換與AI溝通模式。研究結果指出，人與AI的協作展現出交融記憶系統的特質，學生主動引導資訊交換以促進知識共享，是啟動有效人—AI協作的關鍵，有助於明確雙方專長分工並達成互補。研究發現人與AI的協作成果能整合於設計任務之中，主要源於迭代的校正行為與人—AI之間的溝通。在學生引導下，AI工具可靈活轉換多種專長角色，挑戰交融記憶系統中知識分工相對固定的認定。從知識維度觀之，概念性知識的擴展能協助學生扮演引導者與資訊轉譯者；但涉及情境脈絡與手感操作的隱性程序性知識則難以被AI取代，彰顯人與AI在知識類型上的互補性。面對人與AI協作的發展趨勢，本研究建議在教學中鼓勵學生主動彙整資料，以建構人與AI之間的知識共享基礎，並融入驗證環節，強化概念性知識培養，同時深化隱性程序性知識的累積，以促進具深度與創造力的人—AI協同共創。

This study investigates collaborative processes and associated knowledge types between design students and generative artificial intelligence (GAI), utilizing theoretical lenses of Transactive Memory Systems (TMS) and knowledge dimensions. Employing a diary method, three design students documented their interactions and collaboration with AI during design tasks, allowing for detailed analysis of task specialization, information exchange, and communication patterns between human and AI members. The findings reveal that human-AI collaboration embodies the characteristics of a TMS, with students proactively facilitating information exchange and promoting knowledge sharing as essential mechanisms underpinning effective collaboration and role clarification. Iterative feedback and communication between students and AI further enabled the integration of collaborative outcomes into design tasks. Guided by students, the AI demonstrated flexibility in performing various specialized roles, challenging traditional TMS perspectives that view knowledge specialization as relatively stable. From a knowledge dimension standpoint, the expansion of conceptual knowledge supported students in acting as facilitators and knowledge translators. In contrast, tacit procedural knowledge, encompassing contextual understanding and manual operations, remained resistant to substitution by AI, emphasizing the complementary relationship between human and AI knowledge. Given the increasing prevalence of human-AI collaboration, this study suggests integrating student-driven data collection and knowledge synthesis within design education to foster a shared knowledge base between humans and AI. Moreover, incorporating verification processes, enhancing conceptual knowledge development, and continually cultivating tacit procedural knowledge are recommended as critical strategies to deepen human-AI collaborative innovation.