Dubbing artificial intelligence in scaffolding

Abstract

This systematic review examines how artificial intelligence (AI) aligns with Vygotsky’s sociocultural framework to enhance ESL/EFL pedagogy through integrated feedback models combining AI-mediated input, peer scaffolding, and Gibbs’ guided self-reflection. Thematic synthesis of empirical literature (2015–2026) across Asia-Pacific contexts reveals that AI effectively scaffolds learners within their Zone of Proximal Development, improving peer feedback quality, self-regulation, and writing proficiency. However, challenges persist, including AI overreliance, variable feedback quality across proficiency levels, and tensions between automated efficiency and human-mediated dialog. The findings indicate that theoretical alignment with sociocultural principles is robust, yet pedagogical efficacy requires a deliberate instructional design that preserves relational engagement and positions AI as a complementary mediator. Successful implementation demands targeted educator training in AI literacy and ethically grounded integration strategies. Future research should prioritize longitudinal designs and cross-contextual investigations to develop scalable, equitable frameworks for autonomous language learning.