Harness the Generative AI tools for Automated Feedback in EAP Writing Assessment

Abstract

This study explores the impact of five generative AI tools (Grammarly AI Plus, XIPU AI, Kimi, Prepostseo, Magic School AI) on English for Academic Purposes (EAP) writing assessment at a Sino-British EMI university. Thirty-seven first-year students wrote 250-word drafts, received AI-generated feedback, and submitted revised versions. Study 1 employed surveys and interviews, revealing that students valued the AI tools for their speed and grammar accuracy. Study 2 analyzed draft and revision scores in Excel, showing improvements in task response and essay structure. However, gains in grammar and vocabulary were limited, which contrasted with student perceptions. The findings indicate that AI feedback can effectively supplement traditional assessment methods, supporting learner autonomy and enhancing certain aspects of writing. Overall, the study highlights the potential of AI tools to aid EAP writing development, while also emphasizing the need to understand their limitations in improving language accuracy.