與GAI合作論證：以內容分析探討臺灣高中生與生成式人工智慧合作模式

Content Analysis of the Collaboration Models Between Taiwanese High School Students and Generative Artificial Intelligence

對生成式人工智慧（Generative Artificial Intelligence, GAI）進行提問，經常被視為一種獲取資訊的途徑。研究人員認為，透過GAI來取得有用資料的學習模式與學生學習表現間的關係是值得探討的。因此，本研究進行了一個實證研究，探討在科學論證的環境中，讓GAI參與學生論證內容蒐集與組織的過程，對於學生的論證表現是否有所影響。本研究在一門科學課程中進行研究。在教學過程，教師會指派科學論證寫作任務，鼓勵學生向GAI提問。學生運用GAI回答的內容進行論證內容的組織，並完成科學論證寫作。本研究蒐集學生的首次和最後一次的科學論證寫作，以及學生向GAI提問的完整對話內容，並運用這些數據進行以下成效探討，包含論證寫作表現、學生向GAI提問的問題類型，以及論證寫作表現與學生提問類型之間的關係。結果顯示，學生活動前與後的論證寫作表現顯著提升，尤其在主張以及理由兩個面向有顯著進步。此外，根據問題類型統計，學生經常對GAI提出的問題類型為事實以及解釋，顯示學生通常向GAI索取事實性的內容。最後，在論證寫作表現以及問題類型的相關性檢定中，發現範例、鞏固以及遺漏的問題類型與學生的論證寫作表現較為相關。根據這項結果，本研究發現學生對GAI作為資訊獲取工具的看法以及GAI作為學生論證寫作學習工具的貢獻與挑戰。

Posing questions to Generative Artificial Intelligence (GAI) is a method for obtaining information. Researchers believe the relationship between this interaction mode and students’ learning performance is worth exploring. Therefore, this study conducted empirical research to investigate whether involving GAI in searching and organizing argumentative information within a scientific argumentation environment affects students’ argumentative performance. For this purpose, data collection and analysis were conducted in a science course; during the teaching activities, teachers assigned scientific argumentative writing tasks and encouraged students to pose questions to GAI, utilizing the content provided by GAI to organize their argumentative content, ultimately completing the scientific argumentative writing. The study collected students’ first and last scientific argumentative writings and their complete dialogues with GAI. This research analyzed the students’ argumentative writing performance before and after the learning activities, the questions posed to GAI, and the relationship between argumentative writing performance and question types. The results showed significant improvements in students’ argumentative writing performance before and after the activities, especially in claims and reasons. Additionally, based on the statistics of question types, the questions frequently posed to GAI by students were factual and explanatory, indicating that students generally seek factual content from GAI. Finally, in the correlation analysis between argumentative writing performance and question types, it was found that example, consolidation, and omission question types were more related to students’ argumentative writing performance. Based on these results, this study highlights the students’ views on GAI as an information acquisition tool and the contributions and challenges of GAI as a learning tool for students’ argumentative writing.