Academic writing in the AI era: enhancing EFL learners' academic writing skills through ChatGPT, critical thinking and academic self-efficacy

Purpose As an effective and valuable language-learning tool, ChatGPT provides dynamic learning experiences. The advent of ChatGPT has revolutionized traditional language-learning methods, transforming them into innovative systems that provide multilingual support and personalized learning experiences. The present study assessed the impact of ChatGPT on the academic writing skills of English as a foreign language (EFL) learners, focusing on paragraph organization, grammar, and sentence structure. Design/methodology/approach A quantitative approach was used to collect data from 325 Malaysian EFL learners via a self-administered questionnaire, using the verified scales for each construct. AMOS version 24.0 was used to perform structural equation modeling (SEM), a multivariate data analysis technique, and path analysis to evaluate the direct and indirect effects of the variables. Bootstrapping approaches were employed to test the mediation and moderation hypotheses using 5,000 samples and a 95% confidence interval. Findings Using ChatGPT significantly enhances the accuracy and validity of EFL learners' academic writing proficiency. Moreover, the present study primarily focused on the role of critical thinking as a mediator, highlighting its function in minimizing cognitive offloading and encouraging more in-depth, reflective learning when using ChatGPT. Additionally, academic self-efficacy significantly moderates ChatGPT's impact on critical thinking and academic writing skills, highlighting the importance of psychological factors in boosting students' confidence and perseverance in AI-supported academic writing activities. Originality/value The originality of this study lies in its use of a mediated-moderated paradigm that has not been extensively examined in previous research on the effects of generative AI, notably ChatGPT, on EFL students' academic writing abilities. This research offers a deeper understanding of the internal mental and psychological mechanisms that influence writing development by examining the mediating role of critical thinking ability and the moderating effect of academic self-efficacy. This study fills a significant gap in the field of language education and AI-assisted learning by offering that students actively evaluate, analyze, and improve AI-generated information rather than merely accepting it, which goes beyond a superficial assessment of AI use and offers empirical data on how distinctive learners' characteristics interact with AI tools to influence writing outcomes. This method encourages students to acquire autonomous writing abilities in addition to AI help by moving the emphasis from tool usage to cognitive engagement. This shows their active and independent learning and cognitive engagement instead of passive dependency.