Exploring university students' acceptance of ChatGPT as a writing assistant : a structural equation modeling approach

Abstract

This study employed Structural Equation Modelling (SEM) and the Technology Acceptance Model (TAM) to investigate university students’ perception of utilizing ChatGPT in the planning and revision stages of writing. Previous research has shown conflicting results, and the proposed TAM-based models lack explanatory power as they neglect students' concerns and the negative utility arising from them, particularly regarding generative artificial intelligence(GenAI).This study proposed an extended TAM by incorporating an external concerns construct, including concerns about plagiarism, accuracy, transparency, and holistic development, that accounts for both positive and negative utility. A key finding indicates that the traditional TAM model is generally effective in explaining ChatGPT acceptance during both the planning and revision stages. Moreover, the study also confirms that the concerns construct plays a role in ChatGPT acceptance within the writing context. This study contributes to understanding the factors influencing students’ intention to adopt ChatGPT and provides evidence for pedagogical approaches, as well as informs educational policies. This study proposed an extended TAM for further research, enabling the exploration of more complex interplay between concerns and TAM constructs as well as shedding light on the direction of evaluating the mediation effects between those constructs. Future research is suggested to validate the extended model across different demographics with a larger sample size.