本研究採行動研究法，建構「基於運算思維之生成式人工智慧教案設計模式」，透過欄位標籤化輸入與四階段分段式生成流程，引導ChatGPT產出具結構性與教學可操作性的教案。本研究提出一套可由工程師與教師複製操作、具規範性與可追溯性的生成式人工智慧教案生成標準化流程，並初步驗證其在企業訓練與教學中的可行性。本研究共進行兩行動循環，分別檢驗企業與教育場域中的教學成效。結果顯示，學習者在運算思維理解、專業內容掌握及教案生成流程理解上皆有明顯進步，企業場域中部分學習者的後測專業表現甚至接近或超越資深工程師參考組；然而，第二輪教育場域實驗發現，學習者容易過度連結運算思維與AI或科技工具，而產生工具化誤解。綜言之，本模式證實生成式人工智慧可輔助教案標準化與訓練傳承，惟仍須澄清運算思維本質。

This study employs an action research methodology to construct a computational thinking-based generative AI lesson plan (GAI-CTLP) design model. Through field-tagged input and a four-stage segmented generation process, it guides ChatGPT to produce lesson plans that are both structured and pedagogically actionable. This study proposes a standard operating procedure (SOP) for generating GAI lesson plans that is replicable by engineers and teachers, and is both prescriptive and traceable. Its feasibility was preliminarily validated in both educational and corporate training settings. There were two action cycles using GAI-CTLP in a corporate training and an educational setting. Results showed that, in general, learners demonstrated significant improvements in CT comprehension, mastery of professional content, and understanding the generation process of GAI-CTLP. In the corporate training setting, some post-test professional performance metrics even approached or exceeded those of the senior engineer comparison group. However, in the educational setting, learners tended to over-associate CT with AI or technological tools, leading to an instrumental misconception. Overall, the proposed model confirms that GAI can support the standardization of lesson plans to help transfer vocational training, yet it is imperative to clarify the fundamentals of CT in doing so.