AI Insight
This mixed-methods study of 486 university students learning English as a foreign language found that using AI-generated feedback for writing was positively associated with deeper cognitive engagement, both directly and indirectly through increased learning autonomy and self-regulated learning behaviors. Students with higher AI literacy showed a stronger relationship between AI feedback use and autonomous learning. Qualitative interviews revealed students engaged in critical evaluation of AI feedback and demonstrated awareness of AI limitations, though approaches varied from rapid surface-level corrections to careful analytical integration.
Why it matters
The findings suggest that simply providing AI feedback tools is insufficient; educational effectiveness depends on students' ability to critically evaluate and strategically regulate AI-generated suggestions. This highlights the importance of developing AI literacy alongside writing instruction to help students use generative AI as a learning tool rather than a passive correction service.
Understand the Science
Generative artificial intelligence (AI) is increasingly used to obtain feedback in English as a foreign language (EFL) writing, but its educational significance depends on how learners interpret and regulate that feedback. Grounded in Self-Determination Theory and self-regulated learning theory, this explanatory sequential mixed-methods study examined an AI-supported motivated self-regulated learning framework using a cross-sectional survey of 486 university EFL students and follow-up interviews with 20 students. Structural equation modeling identified a positive direct association between reported AI feedback use and deep cognitive engagement, together with a positive indirect statistical association through learning autonomy and self-regulated learning. The AI feedback use–autonomy association was stronger at higher modeled levels of AI literacy. Interview accounts described feedback-related control, critical evaluation, and awareness of AI limitations, alongside contrasting experiences of rapid polishing and careful evaluation despite limited familiarity with AI. Integration of the two strands highlighted variation in how learners evaluated and incorporated feedback. These self-report findings do not establish temporal ordering or causal effects. They identify learner agency, strategic regulation, and critical evaluation as relevant considerations for future investigations of AI-supported writing.