Charting Uncharted Waters: AI-Driven Feedback and Writing Accuracy of Intermediate Iranian EFL Learners

This study explored the impact of AI-driven feedback on the writing accuracy of Iranian intermediate EFL learners using a quasi-experimental design. Through convenience sampling, 100 participants (male and female, aged 18-25) from two language institutes in Tehran were divided into four groups: two experimental groups receiving direct and indirect feedback from AI ChatGPT, and two control groups receiving the same feedback types from their teacher. Participants completed a pre-test, ten writing tasks over 14 weeks, and a post-test. Results, analyzed via descriptive statistics and one-way ANOVA, showed significant improvements in writing accuracy across all groups. The AI direct feedback group achieved the highest improvement, with a mean difference of 7.34 (p < 0.05). Post hoc comparisons revealed that AI direct feedback significantly outperformed both AI indirect feedback and traditional teacher feedback, while teacher indirect feedback resulted in the least progress. Error-type analysis highlighted the superior effectiveness of AI feedback, particularly direct feedback, in reducing spelling and verb form errors. These findings highlight the potential of AI-driven feedback-both direct and indirect-to exceed traditional teacher feedback methods in improving EFL learners’ writing accuracy, with direct feedback emerging as the most effective approach.