生成性AI在寫作評估中的研究綜述：機遇、挑戰與研究展望

A Synthesis of Research on Generative AI in Automated Writing Evaluation: Opportunities, Challenges, and Future Directions

自2022年11月ChatGPT發佈以來，生成式人工智慧（GenAI）的快速發展已在語言教學與學習的各個層面，特別是在寫作評估方面，引發了巨大的變革。現有研究已探討將GenAI工具應用於寫作評估中的可能性，但其結論仍存在一定爭議，這在一定程度上限制了我們對GenAI工具在寫作評估中，尤其是在寫作評分以及提供寫作回饋方面之有效性的理解。因此，有必要對現有研究進行整合性綜述，以進一步釐清GenAI工具在寫作評估中的角色與成效。本綜述採用主題分析法（Braun & Clarke, 2006），整合分析了2022年至2024年間的18項實證研究，並歸納出三個主要主題，包括：（a）GenAI工具在作文評分中的有效性；（b）GenAI工具在提供書面回饋方面的能力；（c）GenAI工具的侷限性與所引發的顧慮。最後，本綜述亦探討未來GenAI工具（例如ChatGPT）在協助教師進行寫作評分與提供寫作回饋方面的潛在可能性，並提出GenAI於寫作評估領域中未來可行的研究方向與教學啟示。

Since the release of ChatGPT in November 2022, the rapid development of generative artificial intelligence (GenAI) has transformed various aspects of language teaching and learning, particularly in automated writing evaluation. Numerous studies have explored the potential of integrating GenAI tools into automated writing evaluation; however, their findings have been somewhat inconsistent. This variability makes it difficult for us to understand GenAI's capabilities in essay scoring and feedback provision within language education. Consequently, a comprehensive and up-to-date synthesis of the existing literature is essential to clarify the role and effectiveness of GenAI in this context.
Using thematic analysis (Braun & Clarke, 2006), this synthesis reviewed 18 empirical studies conducted between 2022 and 2024, identifying three key themes: (a) the effectiveness of GenAI tools in essay scoring, (b) the capabilities of GenAI tools in providing written feedback, and (c) the limitations and concerns associated with their use. This review evaluates the potential applications of GenAI tools, such as ChatGPT, in supporting teachers with automated essay scoring and feedback delivery. Additionally, it highlights emerging research directions and pedagogical implications for integrating GenAI into automated writing evaluation practices.