Perceived Intelligence and Perceived Anthropomorphism as Additional Predictors of ChatGPT Usage Behaviour

Abstract

The rapid advancement of generative artificial intelligence (AI) tools like ChatGPT is transforming various sectors, including education. Although many are adopting ChatGPT for learning, not all students are embracing it. This is because there are contrasting views regarding its usefulness, ease of use, reliability, and ethical implications. ChatGPT acceptance and usage behaviour have been frequently studied through the Unified Theory of Acceptance and Use of Technology (UTAUT) model. However, it may not completely describe the features of AI. Hence, this paper extends the UTAUT model by incorporating two constructs specifically on AI, namely, Perceived Intelligence and Perceived Anthropomorphism as predictors of the ChatGPT usage behaviour. The study followed a quantitative research design. Purposive sampling was used to collect data from 384 students at a public university in Northern Malaysia. Data was analyzed through Partial Least Squares Structural Equation Modeling. The results showed that all six predictors have significant effects on the Behavioural Intention with 48.3% of variance. Behavioural Intention predicts the actual Use Behaviour with 16.5% of variance. The results highlight the robustness of the extended UTAUT model, where the two constructs become predictors of behavioural intention to use. This indicated students’ dependency on the perceived reasoning of ChatGPT. Based on these predictors, the study has constructed a behavioural model to determine the ChatGPT usage among university students. The findings contribute by demonstrating the validity of the extended UTAUT model. Additionally, practical recommendations are provided to universities, educators, and AI developers for integrating ChatGPT in academic settings.