ChatGPT, a generative artificial intelligence (GAI) tool, has gained recognition as a valuable educational technology (EdTech) resource, yet its impact on students' academic outcomes in Ghana remains unclear. This study examined the impact of use of ChatGPT on academic achievement among higher education (HE) students using the “interactive-constructive-active–passive (ICAP)” framework and “self-determination theory (SDT)”. A purposive sample of 457 HE economics students from a public university in Ghana was selected. The study used adapted scales of ICAP (i.e. measuring the use of ChatGPT), “desire to learn (DFL), self-efficacy (SE), beliefs about the future (BAF), epistemic curiosity (EP) and perceived academic performance (AP)”. Data were analysed using “partial least squares structural equation modelling (PLS-SEM) and fuzzy-set qualitative comparative analysis (fsQCA)”. PLS-SEM results indicated that the use of ChatGPT (ICAP) positively influenced DFL, SE, BAF and AP. DFL and BAF were positively associated with AP, whereas SE showed a negative association with AP. Furthermore, DFL, SE and BAF mediated the relationship between the use of ChatGPT and AP. EP directly influenced DFL, SE and BAF and moderated the positive influence of ChatGPT usage on DFL and BAF. FsQCA results revealed four configurations for achieving high AP. This study highlights the positive role of ChatGPT in the academic performance of economics students and underlines the importance of considering multiple factors in policy and intervention strategies to optimise the use of AI in higher education.