<![CDATA[The rapid integration of Generative Artificial Intelligence (GenAI) in higher education has sparked global concerns regarding academic integrity and regulation. However, existing literature often generalizes student perceptions through a binary STEM versus non-STEM lens, obscuring crucial intradisciplinary variations. This study investigated the disparities in ethical awareness and regulatory demands regarding GenAI among university students, specifically challenging the monolithic categorization of Non-Exact Sciences. Employing a quantitative cross-sectional survey, data from 721 Indonesian undergraduates were collected using a 25-item Likert-scale questionnaire, validated via Pearson correlation with high reliability (Cronbach’s α = 0.957). Statistical analysis using the Kruskal-Wallis H test revealed a significant divergence from traditional assumptions. While macro-analysis confirmed Exact Science students generally exhibit higher ethical awareness, granular post-hoc analysis uncovered Language Education students represent a positive anomaly. These findings empirically operationalize the concept of 'epistemological proximity,' the degree to which GenAI intersects with a discipline's core ontology, such as text production in language studies. As a measurable predictor of AI ethical awareness, students in text-centric disciplines (Language Education) demonstrated significantly higher ethical (Mean Rank = 419.64) and regulatory demand scores (Mean Rank = 403.17) compared to other Non-Exact Sciences (p < .001), rendering their profiles statistically indistinguishable from Exact Sciences. Thus, epistemological proximity is a stronger and more operational predictor of ethical awareness than broad binary disciplinary labels. The study concludes that the non-exact sciences category is no longer relevant for AI adoption studies and recommending policymakers develop discipline-sensitive AI literacy programs addressing the specific regulatory vacuums.]]>
Copyrights © 2026
