AI-driven pathways to improve writing performance of ESL learners to drive innovation and sustainability in higher education

In Pakistan, students of English as Second Language (ESL) often struggle with writing performance (WP) due to limited self-efficacy, insufficient feedback, and lack of personalized support. Traditional classroom practices provide limited opportunities for sustained writing improvement. With growing reliance on technology, the integration of AI tools (e.g., Auto-Peer and other automated feedback systems) offers new possibilities; however, their role in enhancing WP remains underexplored. Therefore, the objective of this study is to examine how AI tools enhance the WP of Pakistani ESL learners by exploring the roles of AI usefulness for sustainable writing, sustainable AI mindset, perceived AI feedback quality, writing self-efficacy (WSE), and transparency in AI. The mediating role of WSE and moderating role of transparency in AI was examined. By considering the cross-sectional research design, data were collected from Pakistani ESL learners from universities through structured questionnaire. In total 277 questionnaires were returned, and data analysis was carried out by using Partial Least Squares Structural Equation Modeling (PLS-SEM). Results suggest that AI tools have a potential role to improve WP of ESL learners. AI usefulness for sustainable writing, sustainable AI mindset and perceived AI feedback quality improve the WP. Furthermore, transparency in AI strengthens the relationship between WSE and WP. These results have valuable insights for policymakers while making strategies to improve WP of ESL learners.