How do ChatGPT's benefit–risk-coping paradoxes impact higher education in Taiwan and Indonesia?

The integration of ChatGPT into higher education in Taiwan and Indonesia presents both opportunities and challenges. This integration creates a paradox of benefits and risks that must be carefully managed. While previous studies have explored ChatGPT's applications, its complexities in educational contexts remain partially unaddressed. This study bridges that gap by integrating the Unified Theory of Acceptance and Use of Technology (UTAUT) with Protection Motivation Theory (PMT) to examine ChatGPT's role through a benefit–risk–coping mechanism. Data were collected from higher education users in Taiwan and Indonesia. Structural Equation Modeling (SEM) revealed distinct patterns in the two regions. In Taiwan, perceived severity reduces the intention to use ChatGPT, while self-efficacy fosters adoption. In Indonesia, users emphasize response efficacy and performance expectancy as stronger predictors of usage intention. Task efficiency and performance expectancy enhance usage intention in both settings, with Indonesia showing a stronger link between intention and actual use. Fuzzy sets Qualitative Comparative Analysis (fsQCA) further identifies diverse configurations for actual usage and disusage of ChatGPT. Task efficiency and performance expectancy emerge as key usage drivers in both contexts. Disusage in Taiwan primarily stems from task inefficiency, whereas multiple factors—including low self-efficacy—hinder adoption in Indonesia. These findings provide practical insights for higher education institutions, guiding strategies to optimize ChatGPT's benefits, minimize risks, and ensure its responsible use in educational settings across Taiwan and Indonesia.