Integrating Ai Writing Assistants into EFL Composition Courses: Effects on Writing Quality and Learner Confidence
Abstract
This paper examines the effects of integrating AI writing assistant tools into a university-level EFL composition course on students' writing quality and self-confidence as writers. A quasi-experimental study was conducted with 148 undergraduate students at New Uzbekistan University over 14 weeks. The experimental group (n=74) received instruction that embedded AI writing assistants — used for planning, drafting, and revision — alongside explicit strategy instruction; the control group (n=74) followed a conventional process-writing approach without AI support. Writing quality was assessed using an analytic rubric across four criteria: task achievement, coherence and cohesion, lexical resource, and grammatical range and accuracy. Learner confidence was measured using a validated self-efficacy writing scale administered at weeks 1, 7, and 14. Results showed statistically significant advantages for the experimental group across all four writing criteria (p < .05), with the greatest gains observed in coherence and cohesion (Δ = 0.79) and task achievement (Δ = 0.77). The experimental group also demonstrated substantially steeper growth in writing self-efficacy across the semester. Qualitative analysis of learner reflection journals highlighted the particular value of AI-generated structural templates, iterative feedback on draft coherence, and the reduction of "blank page" anxiety at the planning stage. Implications for writing pedagogy and AI tool design are discussed.