<![CDATA[This study examined the effectiveness of artificial intelligence (AI) feedback systems in improving grammatical accuracy in contextual writing tasks among English as a Foreign Language (EFL) learners. Thirty undergraduate students from twelve different study programs at Universitas Sains Islam Al Mawaddah Warrahmah Kolaka participated in this research as part of a general English course. Using a mixed-methods quasi-experimental design with pretest-posttest control group framework, this study compared the effectiveness of AI-driven feedback (via Grammarly) versus conventional teacher feedback in improving grammatical accuracy within authentic writing contexts. Data were collected through written assignments, error analysis, and student perception interviews. Results demostrated that students receiving AI feedback demonstrated statistically significant improvements in grammatical accuracy (M = 82.5, SD = 6.3) compared to the control group (M = 74.3, SD = 7.1), with p 0.001. Notably, AI systems excelled at detecting and correcting morphosyntactic errors, subject-verb agreement violations, and determiner misuse. Qualitative findings revealed that students perceived AI feedback as immediate, non-judgmental, and conducive to self-correction and engagement. However, students also expressed concerns about over-reliance on automated tools. This research suggested that a hybrid approach integrating AI feedback with traditional instructor guidance offers optimal outcomes for developing grammatical competence in contextual writing. Pedagogical implications include leveraging AI tools as formative assessment instruments alongside explicit grammar instruction and peer review mechanisms.Keywords: AI feedback systems, grammatical accuracy, contextual writing, EFL learners, and automated writing evaluation.]]>
