대학생의 생성형 AI 학업 활용에서 의존/부담과 규칙 명확성 요구의 윤리 불안 예측 요인 분석

초록

This study explores predictors of ethical anxiety (academic-integrity-related anxiety) among college students in the context of using generative AI for academic work. In particular, it examines how experiences of dependence/burden during learning and the demand for rule (criteria) clarity at the course and institutional levels are associated with ethical anxiety. Using a convergent mixed-methods design, an online survey (Likert-scale items and open-ended responses) was administered to 131 undergraduates at D University in December 2025. Quantitative data were analyzed using descriptive statistics, Spearman correlations, and multiple regression with interaction terms, and qualitative responses were supplemented through thematic analysis. The results show that students frequently use generative AI across a range of academic tasks and report reduced academic stress; however, cognitive expansion was accompanied by salient experiences of dependence/burden. Ethical anxiety was more strongly related to dependence/burden than to usage frequency, and mere awareness of rules or guidelines did not sufficiently account for ethical anxiety. By contrast, a stronger demand for specific criteria in “partial allowance” situations was positively associated with ethical anxiety, and a moderating effect indicated that the impact of dependence/burden on ethical anxiety varied by the level of demand for rule clarity. Perceived fairness (competitive) pressure also exceeded the midpoint. These findings suggest that, rather than reducing generative AI to an allow/ban dichotomy, course and assessment design should be grounded in guidelines that integrate support for learner agency, actionable specification of permissible boundaries, and fair assessment practices.

키워드