AI in academia: How do social influence, self-efficacy, and integrity influence researchers' use of AI models?

Abstract

The integration of artificial intelligence models into academic settings has experienced remarkable growth in recent years. Given that researchers' interactions with and perceptions of these technologies can substantially influence academic procedures and outputs, identifying the key determinants of their incorporation into university environments is crucial. This investigation pursued two main objectives: first, to identify the variables that condition the implementation of AI models in research activities, and second, to analyze how perceived ethical considerations and academic integrity influence their adoption. The empirical study was conducted through a digital survey administered to 302 academic researchers from Peruvian public and private universities. The analytical methodology employed structural equation modeling and confirmatory factor analysis, grounded in an expanded version of the Unified Theory of Acceptance and Use of Technology 2 model. The results demonstrated that six out of nine hypotheses were supported; social influence, educational self-efficacy, and academic integrity were identified as primary factors predicting researchers' use of AI models. Effort expectancy had a significant negative effect on AI model use. Furthermore, the use of AI models was found to significantly influence both teachers' concerns and perceived ethics among academics. Notably, performance expectancy, technological self-efficacy, and personal anxiety did not significantly affect AI model use. This study contributes to the understanding of AI adoption in academic research by highlighting the importance of social, educational, and ethical factors. These findings have implications for developing policies and training programs to promote responsible AI use in higher education and suggest a need to reevaluate traditional technology acceptance models in the context of AI in academia.