Narcissism, Generative Artificial Intelligence Use Orientation, and Academic Dishonesty among Undergraduates at a Nigerian University

Abstract

Generative artificial intelligence (GAI) has changed how academic dishonesty is enacted, yet little is known about whether narcissism relates to dishonesty through GAI use. This cross-sectional study examined whether GAI use orientation statistically accounts for part of the association between narcissism and academic dishonesty among University of Ilorin students. A total of 401 second- to fourth-year students (aged 18–33 years, M = 20.64), were recruited from six faculties through quota and convenience sampling. Participants completed the Brief Pathological Narcissism Inventory, a researcher-developed Generative AI Use Scale, and an adapted McCabe/ICAI Academic Misconduct Inventory with five added AI-specific items. In a hierarchical regression controlling for ethical AI training, narcissism (β =.316, p <.001) and GAI use orientation (β =.154, p =.002) each uniquely predicted dishonesty, explaining 15.1% of the variance. In an exploratory model with the two subscales, vulnerable narcissism (β =.248, p =.001) but not grandiose narcissism (β =.090, p =.210) was a unique predictor, although the two coefficients did not differ significantly (p =.174). The bootstrapped indirect effect of narcissism through GAI use orientation was small but significant for the full outcome (ab = 0.025, 95% CI [0.010, 0.044]; 12% of the total effect) and non-significant once the AI-specific items were removed (ab = 0.006, 95% CI [−0.004, 0.019]), suggesting overlap in content between mediator and outcome. Narcissism remained directly related to dishonesty under both scorings. Longitudinal designs and non-overlapping measures are needed.