Exploring the Potential of ChatGPT 5.0 and LexiBot as Automated Essay Scorers for IELTS Writing Task 2: A Comparison with Human Raters

Abstract

With the rapid growth of artificial intelligence (AI) in education and the increasing demand for accurate and efficient language assessment, this study explores the potential of AI as an automated scoring tool for IELTS Writing Task 2 essays. A total of 74 IELTS Writing Task 2 essays, sourced from official Cambridge IELTS practice tests and previously evaluated by certified IELTS examiners, were re-assessed by ChatGPT and LexiBot using the official IELTS Writing Band Descriptors. Quantitative analyses, including intraclass correlation coefficients (ICC), mean absolute error (MAE), and Bland-Altman analysis, were employed to examine the degree of alignment between AI-generated and human-assigned scores. The results suggest that ChatGPT demonstrated relatively stronger alignment with human ratings than LexiBot, particularly at the individual essay level, and that more recent versions of ChatGPT exhibited improved scoring performance compared with earlier models. However, a tendency toward lower score assignment by AI tools indicates limitations in their ability to fully replicate human judgment. Accordingly, AI systems cannot yet be considered substitutes for certified human raters, but may serve as complementary formative assessment tools to support teachers’ grading and learners’ writing development during IELTS test preparation.