<![CDATA[Academic writing proficiency is crucial for Indonesian university students navigating global careers, yet the high student-to-teacher ratio severely limits the provision of timely, quality feedback. This constraint impedes skill acquisition and necessitates the exploration of scalable technological solutions. This study aimed to evaluate the validity and causal efficacy of a specialized, localized AI-powered Automated Essay Scoring (AES) system in accelerating students’ writing skill development over one academic semester. A quasi-experimental, pretest-posttest control group design (N=120) was used. The experimental group received continuous AI feedback, while the control group received traditional manual feedback. ANCOVA was applied to measure skill gain, supported by qualitative data on faculty workload and user acceptance. Findings showed the AI group achieved superior learning gains (13.5 vs. 5.5 raw gain) with a statistically significant main effect (F=22.45, p < 0.001). The system demonstrated high scoring reliability (r=0.88) and reduced lecturer routine grading time by 65%, successfully driving improvement in higher-order skills like Structural Coherence. The research confirms that the customized AI-Feedback Model is a pedagogically transformative tool that provides a sustainable solution to structural constraints in EFL education. It establishes a new paradigm for instructional practice by positioning localized AI as a highly consistent and effective mechanism for mass academic skill development in resource-constrained educational environments.]]>
Copyrights © 2025
