Trust in ChatGPT and Perceived Academic Writing Improvement: A TAM- Based Quantitative Study in a Pakistani ESL Context

Abstract

This study examines the impact of undergraduate students’ trust in ChatGPT on their perceived improvement in academic writing and their intention to utilise the tool in future writing tasks. Grounded in the Technology Acceptance Model (TAM) and the Trust in Technology Framework, this research employs a quantitative approach to examine student perceptions within a Functional English course at a public-sector university in Pakistan. A total of 225 students from the Telecommunication Engineering, Computer Science, and Chemistry departments completed a structured survey. Using descriptive statistics, Pearson’s correlation, and multiple regression analyses, results demonstrated strong positive correlations between trust in ChatGPT and students’ perceived improvements in clarity, vocabulary, and organisation (r = 0.75, 0.78, and 0.82, respectively; p < 0.001). Furthermore, multiple regression results revealed that trust and technology acceptance were significant predictors of students’ future intention to use ChatGPT (β = 0.41, p < 0.001), collectively explaining 63% of the variance in students’ adoption intentions. The study offers a novel contribution by providing localised insights from a non-Western English as a Second Language (ESL) context, where research on generative AI in academic writing remains scarce. Findings suggest that students’ confidence in AI-generated feedback fosters writing development and highlights the importance of institutional support, teacher training, and the ethical integration of AI into academic practices. Limitations include the study’s reliance on self-reported perceptions, the focus on a single institution, and the exclusion of longitudinal performance measures. Future research should incorporate mixed-method approaches and diverse ESL contexts to enhance generalizability.