Critical Thinking Development in GenAI-Supported EFL Argumentative Writing: A Longitudinal Study

Abstract

The rapid development of generative artificial intelligence (GenAI) has prompted increasing interest in its pedagogical potential in second language (L2) writing. While prior research has predominantly examined outcomes or learner perceptions, considerably less attention has been paid to the developmental processes through which GenAI mediates critical thinking over time. Conceptualizing GenAI as a human–AI co-regulatory partner, this longitudinal qualitative case study investigates how sustained GenAI-supported interaction shapes critical thinking development in EFL argumentative writing. Drawing on co-regulation theory, metacognitive scaffolding, and cognitive views of critical thinking, the study traces the evolving interactional patterns of four Chinese EFL learners across an eight-week instructional cycle. Data sources include human–AI dialogue transcripts, successive writing drafts, reflective journals, and semi-structured interviews. Integrated longitudinal analysis reveals diverse developmental trajectories influenced by learner beliefs about AI, epistemic orientation, and instructional framing. The findings highlight conditions that enable or constrain co-regulation and offer practical insights for designing AI-mediated writing instruction that fosters critical thinking.