Higher education and ChatGPT: A multi-country study on epistemic emotions and their antecedents

Emotions tied to knowledge acquisition and creation are essential for higher education (HE) students in the 21st century as they face the complexities of today’s digital society. According to Pekrun’s generalized control-value theory (CVT), epistemic emotions are strictly interrelated with a range of antecedents and outcomes. Despite the recent surge in research in this field, there is a lack of studies on how students function when faced with generative artificial intelligence (AI). Therefore, we investigated epistemic emotions experienced by HE students when using ChatGPT – the currently leading interactive generative AI system – and their antecedents. Our sample included 11,427 HE students from over 127 countries, who responded to an online survey on ChatGPT use, control, value, and epistemic emotions. A structural equation model indicated positive relations between ChatGPT use and ChatGPT control and value. In turn, ChatGPT control was associated positively with curiosity and negatively with surprise, enjoyment, anxiety, frustration, and boredom; and ChatGPT value was associated positively with positive emotions and anxiety, and negatively with the remaining negative emotions. The model was invariant across gender, study field, and cultural context, confirming the universality of CVT. A linear mixed model revealed a higher frequency of positive rather than negative emotions. It also showed higher enjoyment, surprise, and anxiety, and lower confusion for students from countries with a lower Human Development Index. These findings suggest that fostering epistemic emotions through key antecedents can help create guidelines for the use of generative AI in HE, which could be shared at various policy levels.