Navigating Generative AI in Higher Education: Mining Engineering Students’ Practices, Perceptions, and Language Choices
Abstract
Generative artificial intelligence (GenAI) has rapidly become an everyday academic resource, making students’ actual practices increasingly important for higher-education policy and pedagogy. This small-scale exploratory survey study examines familiarity with, frequency and purposes of use of, attitudes toward, verification practices, and language choices in GenAI among 30 second-year undergraduate students from Mining Engineering, Petroleum Engineering and Environmental Engineering programmes at the Faculty of Mining and Geology University of Belgrade. For the purposes of the study, the term Mining Engineering students is used as a departmental umbrella term encompassing these three distinct study programmes. Data were collected through an anonymous paper-based questionnaire combining closed-ended, multiple-response, Likert-type, and open-ended items. Descriptive statistics were complemented by exploratory respondent-level comparisons using Fisher’s exact tests, Mann–Whitney U tests, and Spearman’s rank correlation where appropriate. Twenty students (66.7%) used AI several times per week or almost every day; 21 (70.0%) used it for solving tasks or problems and 17 (56.7%) for understanding course content. Twenty-five students (83.3%) reported asking follow-up questions to obtain more precise information, while 10 of 28 valid respondents (35.7%) checked AI-generated information often or almost always. Open responses primarily emphasised speed and efficiency as benefits and inaccuracy or unreliability as limitations. Language use was flexible: Serbian was prominent, but English and bilingual interaction were also common. Higher self-reported English proficiency was strongly associated with English-involved AI use in an exploratory Fisher’s exact test (p < .001). The findings suggest pragmatic rather than uncritical GenAI adoption and point to the need for discipline-sensitive AI literacy, transparent institutional guidance, and explicit attention to multilingual AI interaction in higher education.