THE IMPACT OF GEMINI AI ON STUDENTS' WRITING PERFORMANCE

Abstract

The integration of Artificial Intelligence (AI) in education offers transformative potential, yet EFL learners still face significant hurdles in writing proficiency. This study aims to explore the effectiveness of Google Gemini, a multimodal Large Language Model (LLM), in enhancing students' writing performance and overcoming psychological barriers, such as low motivation. Using a qualitative approach grounded in the Technology Acceptance Model (TAM) and Socio-constructivism theory, this research examines how Gemini's real-time feedback functions as a "More Knowledgeable Other" (MKO). The findings demonstrate that Gemini effectively identifies deficiencies in sentence structure and grammar, acting as a vital catalyst for boosting fluency and coherence in EFL compositions. Furthermore, students perceive the tool as highly intuitive, particularly during the initial drafting and planning phases. The study concludes that while Gemini provides essential scaffolding within the Zone of Proximal Development (ZPD), structured pedagogical supervision is necessary to prevent over-dependency. These results suggest that a synergy between AI support and human guidance is fundamental for sustainable second-language mastery in the digital age.