AI錯題診療室—以教育型AI促進國小高年級社會領域個人化學習與差異化補救
講者/Speaker： 高珠容, Kao Chu Jung
本教學應用以「AI 錯題診療室—以教育型 AI 促進國小高年級社會領域個人化學習與差異化補救」為主題，針對國小學生在社會領域學習中常見的概念理解不足、考前複習缺乏個別化檢核，以及考後訂正容易流於抄寫正解等問題，設計以教育型 AI 學習工具 e 度作為對話引導式學習支援。課程以五年級與六年級學生為實施對象，將 AI 融入「平時概念釐清、考前題目化檢核、考後錯題診療」三個學習時點，透過蘇格拉底式追問，引導學生完成「找錯因、釐清概念、修正表達、類題再練習」的學習循環。
本設計強調 AI 並非答案提供者，而是協助學生思考、重述與驗證的學習夥伴。教師透過明確的提問框架與學習證據要求，規範學生須先寫下卡關點，再與 e 度對話，最後以自己的話整理概念並完成類題練習，以降低直接複製 AI 回答的依賴風險。量化結果顯示，在平時使用 e 度的班級中，社會領域評量平均分數小幅提升；而加入考前複習與考後錯題診療的班級，平均成績提升更為明顯，且後段學生進步幅度較大，呈現差異化補救的潛力。學生回饋亦顯示，多數學生能感受到 e 度對理解與解題的幫助，同時也能覺察過度依賴 AI 的風險，並開始形成批判使用與自我監控的學習態度。
整體而言，本教學設計將教育型 AI 從「生成答案」轉化為「生成思考歷程」的輔助工具，透過教師引導、學習證據與錯因診療流程，促進學生概念理解、表達能力與個人化補救學習。此模式具備國小教育現場的可行性，亦可遷移至國語、自然、數學等不同領域，作為推動 AI 素養與差異化教學的實務參考。
This teaching application, titled “AI Error Diagnosis Clinic: Using Educational AI to Promote Personalized Learning and Differentiated Remediation in Upper Elementary Social Studies,” addresses common learning challenges among upper elementary students. These include insufficient conceptual understanding, a lack of individualized review prior to assessments, and post-assessment corrections that often become the mere copying of correct answers. The instructional design utilizes e-Du, an educational AI learning tool, as a dialogue-based learning support system. Implemented with fifth- and sixth-grade students, the approach integrates AI into three key learning stages: daily concept clarification, pre-assessment question-based review, and post-assessment error diagnosis. Through Socratic questioning, students are guided through a learning cycle of identifying error causes, clarifying concepts, revising explanations, and practicing with similar questions.
This design emphasizes that AI should not serve as a direct answer provider, but rather as a learning partner that scaffolds students’ thinking, restatement, and verification. Guided by teacher-designed questioning frameworks and required learning evidence, students first identify their points of confusion, engage in dialogue with e-Du, and finally reorganize concepts in their own words before completing similar practice tasks. This structured process helps reduce the risk of students directly copying AI-generated responses. Quantitative results indicated that the class using e-Du for regular learning support demonstrated a slight improvement in Social Studies assessment scores. In contrast, the class that additionally implemented pre-assessment review and post-assessment error diagnosis showed more significant score gains, particularly among lower-performing students, highlighting the potential of this approach for differentiated remediation. Furthermore, student feedback revealed that most students recognized e-Du’s usefulness in supporting understanding and problem-solving; they also became aware of the risks of overreliance on AI and gradually developed a more critical and self-regulated attitude toward its use.
Overall, this teaching design transforms educational AI from a tool for generating answers into a tool for fostering thinking processes. Through teacher guidance, learning evidence, and structured error diagnosis, it promotes students’ conceptual understanding, expressive ability, and personalized remedial learning. This practical model is adaptable to elementary school settings and other subject areas, such as Mandarin, Science, and Mathematics, serving as a feasible reference for promoting AI literacy and differentiated instruction.
- 論文全文：AI錯題診療室—以教育型AI促進國小高年級社會領域個人化學習與差異化補救
- 影像： 相關影片