利用AI工具協助高中生英文寫作能力之後設認知力、科技創新意識、自我效能、趣味與求知價值及學習表現提升

Abstract

在台灣的高中生面臨學科能力測驗的情況下，英文寫作一直都是在英文學科中帶給學生及現場教師極大挑戰的一部分，教師面臨學生人數眾多、批改壓力繁重、批改標準難統一標準化等困擾，而學生端面臨回饋不即時、寫作興趣不高、後設認知不足、自我效能感受不彰等挑戰。傳統英文寫作教學中常使用的同儕互評亦因學生專業素養不足、人際互動壓力，在現場的執行效果有限。自2022年ChatGPT等AI工具的推出後，各界爭相使用並期望測試其效能，其中不乏各種教育相關的應用。本研究以新竹某高中學生為研究對象，期望透過引入AI工具，改善上述英文寫作的困難點，提供更具趣味和個人化的學習。透過AI工具，本研究期望探討學生的後設認知能力、科技創新意識、AI自我效能、趣味和求知價值，和學習表現之間的相關性。這項研究將對高中英文寫作教學和學生學習帶來有益的啟示，並有望為AI工具在教育領域的應用提供實證基礎。本研究使用實驗研究和問卷調查法，採立意取樣，邀請新竹地區某公立高中學生為研究對象，為期八週。問卷係經參考相關文獻後編制，包含「後設認知力」、「科技創新意識」、「自我效能」、「體驗價值-趣味價值」和「體驗價值-求知價值」量表。透過 SPSS 23與Smart PLS 執行結構方程模式分析及驗證。得到以下研究結果：一、 後設認知力與ChatGPT自我效能呈正相關二、 科技創新意識與ChatGPT自我效能呈正相關三、 ChatGPT自我效能與求知性價值呈正相關四、 ChatGPT自我效能與趣味性價值呈正相關五、 求知性價值與學習表現不具相關性六、 趣味性價值與學習表現呈負相關七、 在差異性分析的結論則得出，原先英文成就分別為低中高之受測者在後設認知力、ChatGPT自我效能、學習表現進步量三構面都有顯著差異Since the introduction of AI tools like ChatGPT in 2022, there has been a widespread adoption in various fields, including education. This study focuses on high school students in Hsinchu, aiming to address the difficulties in English writing by introducing AI tools to provide a more interesting and personalized learning experience. Using AI tools, this study aims to explore the correlation between students' metacognitive abilities, technological innovation awareness, AI self-efficacy, interest, intrinsic value, and learning performance. The findings of this research are expected to provide valuable insights for high school English writing instruction and student learning and contribute empirical evidence to the application of AI tools in the field of education. This study employs experimental research and questionnaire surveys, utilizing purposive sampling to invite high school students from a public school in Hsinchu as research participants over an eight-week period. The questionnaire, developed with reference to relevant literature, includes scales for"metacognitive ability," "technological innovation awareness," "self-efficacy," "experiential value – interest value," and "experiential value – intrinsic value." Structural equation modeling analyses and validations were conducted using SPSS 23, and Smart PLS. The following research results were obtained: 1. Metacognitive ability is positively correlated with ChatGPT self-efficacy. 2. Technological innovation awareness is positively correlated with ChatGPT self-efficacy. 3. ChatGPT self-efficacy is positively correlated with epistemic value.4. ChatGPT self-efficacy is positively correlated with hedonic value.5. Epistemic value is not correlated with learning performance.6. Hedonic value is negatively correlated with learning performance.7. The analysis of differences concluded that based on the participants' initial English performances, differences in the three dimensions of metacognitive ability, ChatGPT self-efficacy, and the improvement of scores are shown.