Generative AI is increasingly used in EFL academic writing, yet its pedagogical value depends on whether students use it as a tool for thinking or merely as a shortcut for text production. This study explored how generative AI prompts functioned as cognitive scaffolding in EFL students’ academic writing, particularly in relation to idea development, writing organization, engagement, and critical thinking. Using a qualitative case study design, the study involved 10 undergraduate EFL students enrolled in a second-semester Academic Writing course at a university in Malang, Indonesia. Data were collected through classroom observations, screen-recorded writing sessions, students’ writing documents, semi-structured interviews, and reflection journals, and were analyzed using reflexive thematic analysis. The findings revealed that generative AI prompts supported students by activating initial ideas, clarifying concepts, and helping them organize writing into more coherent and logical structures. AI prompts also reduced hesitation and increased students’ confidence during the writing process by offering immediate support when they felt stuck. However, the findings also showed a risk of overreliance, as some students copied or minimally revised AI-generated text without sufficient evaluation. More reflective students used AI prompts to compare ideas, identify weaknesses, evaluate paragraph connections, and revise arguments more purposefully. These findings suggest that generative AI prompts can serve as cognitive scaffolding when they encourage planning, evaluation, and reflective revision, but they may weaken students’ intellectual ownership when used passively. To maximize their value in EFL academic writing, AI-assisted instruction should include explicit prompt literacy, critical evaluation of AI responses, revision-based tasks, and teacher-guided reflection so that students write with AI rather than by AI.
Copyrights © 2026