‘Critical chatting’ or ‘casual cheating’: how graduate EFL students utilize ChatGPT for academic writing

With the penetration of generative artificial intelligence (AI) tools like ChatGPT into higher education, critical thinking (CT) is increasingly paramount in students' academic writing. However, little is known about EFL students' application of such tools for academic writing, especially how CT is embedded in this process. Incorporating the CT skill framework and the Four Resources Model of critical literacy, this qualitative case study explored four graduate EFL students' ChatGPT-assisted academic writing, focusing on their purposes, rationales, and CT in this process. Multiple data-collection methods were adopted to form a comprehensive understanding, and thematic analysis was employed to analyze the multi-source data. The findings revealed that participants applied ChatGPT mainly for language editing, knowledge inquiry, and inspiration seeking, underpinned by the rationales including language enhancement, information search efficiency, and broadening perspectives. Additionally, students demonstrated CT in their multifaceted roles as code breakers, text participants, text users, and text analysts, integrating core CT skills in this social literacy practice. In conclusion, students demonstrated a sound understanding of ChatGPT's affordances for academic writing and a tendency of 'critical chatting' rather than 'casual cheating,' despite varying depth. These findings inform future pedagogical strategies and policymaking regarding integrating AI tools in EFL students' academic pursuits.