AI Writing Assistants in English Language Learning: Evaluating Feedback Quality and Learner Autonomy

Abstract

As artificial intelligence (AI) writing assistants become increasingly integrated into English language learning (ELL), their influence on feedback quality and learner autonomy warrants critical evaluation. This mixed-methods study investigates how AI-generated feedback compares to teacher feedback in terms of accuracy, clarity, usefulness, and its impact on learner autonomy. Forty university-level ELLs completed writing tasks using either AI tools or instructor input. Results showed that while AI feedback was effective for correcting surface-level errors, it lacked the pedagogical depth necessary to foster meaningful learning. Teacher feedback, by contrast, encouraged reflective revision, metacognitive engagement, and greater writing independence. Despite AI tools’ convenience and immediacy, learners often accepted suggestions passively, which hindered the development of critical evaluation skills and self-regulated learning. The study concludes that AI writing assistants can serve as useful supplements in writing instruction but should not replace human feedback. Instead, a hybrid model that combines technological efficiency with pedagogical insight may offer the most effective support for developing autonomous, reflective writers.