3 Group scores

This study investigates the impact of two distinct AI-mediated feedback modalities—analytic feedback (Gemini) and conversational feedback (TalkAI)—on EFL learners’ performance in IELTS Writing Task 2, compared with traditional teacher-mediated feedback. Using a randomized pretest–posttest mixed-methods design, 90 Iranian IELTS candidates were assigned to Gemini, TalkAI, or control groups over a six-week instructional period. Writing performance was assessed by certified IELTS examiners, and quantitative analyses (paired-sample t-tests, ANOVA, and ANCOVA) demonstrated that both AI-assisted groups achieved substantial and statistically significant gains, outperforming the teacher-feedback group, with no significant difference between analytic and conversational AI feedback. Qualitative interview data further revealed that learners perceived AI feedback as more immediate, clear, and supportive of revision autonomy and confidence. Together, the findings indicate that both analytic and conversational AI systems can effectively enhance argumentative writing performance in high-stakes IELTS preparation contexts, highlighting their potential role in scalable, equitable writing support.