| dc.description.abstract | 本研究旨在精進「教師專業發展」課程中「真實情境議題研討」的品質，探究 ChatGPT 人機協作的設計，在課室中支援個別化學習與深化思考等方面的作用、價值與限制。研究歷程中先後建構並實施兩版自定義 GPT：「蘇格拉底 1.0」模擬蘇格拉底式提問，引導師資生系統思考；「蘇格拉底 2.0」則轉化為 AI 協作教練，運用七項策略（如積極探問、提示教學、角色扮演、帶入經驗、提供多元選項等），逐步優化議題研討的品質。
透過課堂觀察、訪談、問卷與對話內容分析（共25位師資生參與），發現師資生在學期初多將 ChatGPT 用於查詢與文書輔助，缺乏與AI協思的意識與應用策略。經歷蘇格拉底 1.0 與 2.0 的協作對話後，AI扮演了促進對話、啟發反思的多重角色，能延伸觀點、補充知識、澄清概念並提供個別化支持。期末資料顯示，師資生能展現批判思考，辨識AI優弱點、整合脈絡觀點，並探索因素間的動態關聯，深化對教育議題的系統思考，並增進其專業認同。
本計畫著重教學資源開發與方法創新，實際驗證生成式AI於教育領域的應用與成效。所建構的ChatGPT人機協作模式，能裨益議題研討的歷程品質，亦有助於培養未來教師面對生成式AI所需的數位轉型素養，對素養導向課程整合AI與數位教學助理的設計，提供具體參照與實證基礎。
This study aims to enhance the quality of issue-based inquiry within the “Teacher Professional Development” course by exploring the design, value, and limitations of ChatGPT-based human-AI collaboration in supporting personalized learning and deeper thinking in the classroom. Two iterations of a customized GPT were developed and implemented throughout the study: “Socratic 1.0,” which simulated Socratic questioning to guide pre-service teachers in systems thinking, and “Socratic 2.0,” which was refined into an AI collaboration coach employing seven strategies (e.g., proactive questioning, prompt engineering, role-playing, incorporating personal experience, and offering diverse options) to optimize the quality of issue discussions progressively.
Data were collected through classroom observations, interviews, questionnaires, and analysis of user-AI dialogues (with 25 participating pre-service teachers). Results indicated that, at the beginning of the semester, most students used ChatGPT primarily for information retrieval and writing assistance, lacking awareness of its collaborative thinking potential. After engaging with Socratic 1.0 and 2.0, ChatGPT played multifaceted roles in facilitating dialogue, stimulating reflection, extending perspectives, supplementing knowledge, clarifying concepts, and providing personalized support. By the end of the course, students demonstrated critical thinking, the ability to recognize AI strengths and limitations, integrate contextual perspectives, explore dynamic interrelationships among factors, and deepen systems thinking related to educational issues, contributing also to their sense of professional identity.
This project focused on developing teaching resources and innovative instruction, providing empirical evidence of the effectiveness of generative AI in education. The constructed ChatGPT-human collaboration model not only enhances the quality of issue-based inquiry but also supports the cultivation of digital transformation competencies essential for future teachers. It offers concrete references and research-based insights for integrating AI and digital teaching assistants into competency-based teacher education. | en_US |
