學習程式設計成為必要且重要的存在，但學習者從零建立基礎到實作應用並不容易。生成式AI的出現提供了工具給學習者輔助，然而由於使用上的高度依賴以及生成結果的不正確性，仍會在學習成效以及未來對工具的使用上造成負面影響。以運算思維運用之，透過對問題的拆解，使用者各別對輸入、輸出、處理方式等各細節分別描述，提高生成式AI對問題的理解。本研究提出從案例式學習防止學習者高度依賴生成式AI以及避免其生成結果的不正確性，以過往案例的解題經驗來延伸應用學習者既有的程式觀念，幫助學習者以運算思維的方式使用生成式AI，利用精準的自然語言建構解題方法，提升學習者學習程式設計時的成效。本研究以此方式設計了一套學習系統，並實施為期十三週的實驗，實驗結果顯示，有別於學習者在電腦環境下直接實作並將完整問題給予生成式AI進行問答的學習方法，以本研究所設計的學習系統學習程式設計，在成效上有著更顯著的效果。;Learning programming has become a necessary and important skill, but it is not easy for learners to build a foundation from scratch and apply it in practice. The emergence of generative AI provides a tool to assist learners, but due to high dependency and the inaccuracy of generated results, it can negatively impact learning effectiveness and future use of the tool. By applying computational thinking, users can improve the understanding of generative AI by describing each detail of inputs, outputs, and processing methods through problem decomposition. This study proposes case-based learning to prevent learners from becoming highly dependent on generative AI and to avoid inaccuracies in generated results. By extending the application of learners′ existing programming concepts through problem-solving experiences from past cases, this approach helps learners use generative AI with computational thinking, constructing problem-solving methods with precise natural language to improve their effectiveness in learning programming. This study designed a learning system based on this approach and conducted a thirteen-week experiment. The results showed that, compared to learners directly implementing and querying generative AI in a computer environment, learning programming using the system designed in this study had more significant effects on learning outcomes.
