應用SCAMPER鷹架AI新詩創作：AI自我效能、興趣型好奇心、匱乏型好奇心、學習興趣、AI使用挫折感、頓悟經驗與新詩創作表現之相關研究

Abstract

本研究旨在探討國小高年級學生在AI輔助新詩創作歷程中，運用SCAMPER策略作為鷹架時，各心理變項之間的關聯與作用機制。研究建構九項假設，涵蓋AI自我效能、認知好奇心(興趣型與匱乏型)、學習興趣、AI使用挫折感、頓悟經驗與詩作表現等構面，藉以描繪學生於人機協作創作情境中的心理動力歷程。本研究採準實驗設計與問卷調查法，以新北市與台北市之國小高年級學生為對象，透過創作評量與量化問卷進行資料蒐集，並運用PLS-SEM結構方程模型進行驗證與分析。研究結果顯示：AI自我效能與興趣型好奇心呈現顯著正相關，惟與匱乏型好奇心無顯著關聯；興趣型好奇心可正向預測學習興趣，但與AI使用挫折感無顯著關聯；匱乏型好奇心則與挫折感呈現正相關，卻與學習興趣無顯著關聯。進一步分析亦發現，學習興趣與頓悟經驗、AI使用挫折感與頓悟經驗皆呈現顯著正相關，指出適度的情緒挑戰有助於觸發創意靈感；而頓悟經驗亦與新詩創作表現呈現顯著正相關。本研究結果呼應並延伸相關理論，指出不同類型好奇心對學習動機與AI經驗的差異性影響，並突顯「頓悟經驗」在語文創作表現中所扮演的中介角色。AI輔助下的人機協作不僅提升學生語言創造力，更促成情感與認知交織的學習經驗，展現詩性思維於科技情境下的嶄新可能。This study investigates the psychological mechanisms involved when upper elementary school students engage in AI-assisted poetry writing using the SCAMPER strategy as a scaffold. Nine hypotheses were proposed, covering AI self-efficacy, epistemic curiosity (Interst type Epistemic Curiosity, I type EC and Deprivation type Epistemic Curiosity, D type EC), learning interest, AI usage frustration, aha! experience, and poetry creation performance. A quasi-experimental design and questionnaire survey were conducted with students from Taipei and New Taipei City. Data were collected through poetry assessments and quantitative surveys and analyzed using PLS-SEM. Results showed that AI self-efficacy was positively associated with I type EC, but not with D type EC. I type EC positively predicted learning interest but was unrelated to AI usage frustration, while D type EC correlated positively with AI usage frustration but not with learning interest. Furthermore, both learning interest and AI usage frustration were positively related to aha! experiences, which in turn significantly predicted poetry creation performance. This study extends existing theories by revealing how different types of curiosity influence AI usage and learning motivation. It also highlights the mediating role of insight in creative expression. Human–AI collaboration in this context supports not only linguistic creativity but also emotional and cognitive engagement, demonstrating new possibilities for poetic reasoning in technology-enhanced environments.