Assessing the importance of variables from a revised technology acceptance model for the use of ChatGPT by university students

Students and researchers are increasingly relying on ChatGPT as a tool to search for information, generate content, make translations and offer research support. This study employs nine variables derived from the Technology Acceptance Model to evaluate whether the variables can predict ChatGPT use among university students in the Philippines. To achieve this, a survey was carried out. Data was collected from 340 students across various faculties of a university. The data were analysed with structural equation modelling (SEM). The model predicts that the nine variables that reflect students' perceptions towards ChatGPT, are indirectly or directly related to the students' use of ChatGPT. The SEM analysis showed that students' performance expectancy, effort expectancy and social influence were related to their attitude towards ChatGPT. The students' attitude together with their perception of the reliability and credibility of ChatGPT were related to intention to use ChatGPT. Intention to use ChatGPT together with trust in this tool were directly related to the actual use of ChatGPT. Additionally, artificial neural network (ANN) analysis was employed for further model optimization. The ANN analysis showed that from the nine predictor variables: 'trust', 'intention to use', 'accuracy' and 'attitude' were most strongly related to students' use of ChatGPT. The findings suggest that to enhance its further adoption, university students should be convinced that ChatGPT can be used in a trustworthy and accurate way. • AI tools revolutionize academic and social science research, aiding data analysis and overcoming language barriers. • SEM reveals significant correlations: effort and performance expectancy, attitude and performance, social influence, and accuracy. • Trust, intention to use, and accuracy are most influential in ChatGPT usage via neural networks.