Self-Efficacy and Strategy in AI-Assisted Academic Writing: A Longitudinal Study of Chinese EFL Doctoral Students

• In Event: AERA Roundtable Session Sunday 1:45 pm - Gold 3
  In Roundtable Session: AI in Student Learning: Programming, Writing, and Engagement

Abstract

As AI tools enter academic writing classrooms, little is known about how EFL doctoral students adapt their self-beliefs and strategies. Grounded in Social Cognitive Theory and self-regulated learning, this mixed-methods longitudinal study examines changes in writing self-efficacy (WSE), revision self-efficacy (Rev-SE), and AI self-efficacy (AI-SE), alongside self-regulated writing strategies (SRWS). Thirty-two Chinese doctoral students completed a 16-week AI-supported writing course. Survey, log, and interview data revealed that WSE improved gradually, especially in grammar and organization; Rev-SE remained low for higher-order revisions; AI-SE increased early but plateaued. SRWS use varied across learners. Three learner profiles—AI-Augmenters, Reflective Integrators, and Cautious Skeptics—emerged. Findings highlight the need to promote strategic awareness and reflective engagement to support writers’ agency in AI-mediated writing contexts.

Authors

• Huanqing Li, Tsinghua University

• Qian Guo, Tsinghua University

• Yanyu Wang, Tsinghua University