Beyond red ink: AI feedback and ESL academic writing development in higher education

Abstract

This study examines the effect of artificial intelligence (AI) generated feedback on the academic writing skills of English as a Second Language (ESL) students at the Faculty of Computing, Islamia University of Bahawalpur, Pakistan. A quasi-experimental mixed-methods design was employed with 400 students from first to eighth semesters, divided into an experimental group (receiving AI-generated feedback using Grammarly and ChatGPT) and a control group (receiving Teacher feedback). Pre and post-test essays, questionnaires, and interviews were used for data collection. Results showed that students receiving AI-generated feedback achieved greater improvements in grammar (43.5%), vocabulary (37.3%), cohesion (49.5%), and overall writing quality (37.8%) compared to the control group, with large effect sizes across all components. Significant reductions were observed in minor errors, including article usage, spelling, and consistency of tense. Qualitative findings revealed that students had positive perceptions of AI-generated feedback, reporting increased awareness of errors, improved confidence, and greater independence during revision; however, some concerns about overreliance were also noted. The study concludes that AI-generated feedback effectively supplements Teacher feedback, enhancing short-term writing performance and indicating potential for increased learner autonomy in ESL academic contexts.