Key points are not available for this paper at this time.
Purpose The application of generative artificial intelligence is on the rise at a very rapid rate among students both in developing and developed countries. This study examines the reasons behind the acceptance and use of ChatGPT among students in developing nations and the gaps in the education levels, with the case of Bangladesh. Design/methodology/approach Data were collected from 307 students using a structured questionnaire. The analysis employed exploratory factor analysis (EFA) to determine the determinants of ChatGPT adoption and its pattern matrix, ordered logistic regression to assess the factors that affect the actual utilization of ChatGPT, and the Blinder–Oaxaca decomposition method to analyze the educational attainment disparity. Findings Findings have shown that performance expectancy, effort expectancy, facilitating conditions, self-efficacy and technological awareness have a significant impact on the actual use of ChatGPT by students. There is a moderating relationship between performance expectancy, effort expectancy, self-efficacy and usage behavior by age. Moreover, there is an educational gap: ChatGPT is used more by students with higher education than by students with secondary education. Practical implications The results indicate that special efforts are required in an attempt to boost digital literacy and equal access to AI-based learning at all levels of education in developing economies. Originality/value This paper combines the UTAUT model with psychological and awareness-based constructs, including self-efficacy and technological awareness. It boosts the theoretical foundation of AI adoption through the amplification of the UTAUT model with demographic moderators and educational level gaps, which offers a solid basis for future research on the emergent AI technology in the educational context. It also adds practical understanding to the students and learning institutions because it provides empirical knowledge on advancing sustainable AI-based learning environments.
Akter et al. (Thu,) studied this question.