Measuring attitudes of ethical use of GenAI in the work of university educators in Slovakia

Abstract

Existing AI ethics frameworks provide broad principles for responsible AI use in higher education. However, empirical evidence remains limited on how university educators prioritize and differentiate concrete ethical dilemmas arising across their academic work. Within the literature included in a preceding systematic literature review, no study empirically examined the complete set of 31 identified ethical problems across seven domains of university educators' academic activity. This cross-sectional quantitative survey therefore investigated educators’ attitudes toward these problems and established their relative priorities for institutional AI governance. Data were collected through an online questionnaire from 388 educators at Slovak public and private universities using a faculty-stratified recruitment strategy. Responses were analyzed using two-tailed one-sample t -tests against the scale midpoint, Cohen's d, repeated-measures ANOVA, and Holm-adjusted pairwise comparisons. Parallel non-parametric analyses were conducted as robustness checks. The analyses were performed in Python 3.14.6 in the Google Colab environment. Mean responses favored ethically safer alternatives for all 31 dilemmas (all p  < 0.001; |d| = 0.36–2.02). Significant differences between dilemmas were found within every academic domain (all p  < 0.001; η 2  = 0.032–0.194), while post hoc analyses identified both comparable subsets and distinct ethical concerns. These findings demonstrate that educators' concerns constitute a differentiated hierarchy rather than a homogeneous set. The study provides an empirically grounded basis for universities, policymakers, and ethics committees to develop ethical codes for AI use, prioritize AI-governance provisions, address comparable dilemmas through shared rules, and develop dedicated principles or implementation guidance for conceptually distinct concerns.