Author: Shamuratova, Dilnoza Jamolovna; Eshmuratova, Aysuliw Rustamovna

Annotation: English as a Foreign Language (EFL) learners often have ongoing fluency issues in self-directed learning settings, which are made worse by restricted access to native speakers. This article examines how Google Gemini, a multimodal Generative AI (GenAI) platform, can help overcome these challenges. Speaking performance could be revolutionized by recent developments in Gemini's capacity to mimic human-like interactions and adjust to individual demands (Widodo et al., 2026). This paper summarizes recent research on how well Gemini supports self-directed practice and task-based learning. According to quantitative data, speaking fluency has improved by an average of 13%, particularly in the areas of lexical diversity and decreased hesitation (Widodo et al., 2026). Increased speaker confidence and the advantages of a nonjudgmental, round-the-clock atmosphere are highlighted via qualitative analysis. The article concludes by examining the methodological challenges of integrating GenAI into formal curricula and the necessity for longitudinal research to substantiate long-term efficacy.

Keywords: Google Gemini, Speaking Fluency, EFL Pedagogy, Task-Based Learning (TBL), Generative AI, Oral Proficiency.

Pages in journal: 755 - 763