Effects of an intervention combining warm therapy with a digital distraction app on pain, stress, and satisfaction during intravenous catheterization in South Korea: a randomized controlled trial.

The rapid diffusion of artificial intelligence (AI) is reshaping university students’ learning processes and modes of academic production (). Across tasks such as writing, paraphrasing, code generation, information retrieval, and synthesis, AI has evolved from a supplementary tool into an embedded component of routine learning environments (; ). While this transformation enhances efficiency and resource accessibility, it also restructures the risk landscape of academic integrity in higher education. Controversies increasingly arise in gray areas where AI use is technically feasible yet normatively ambiguous, shifting concerns away from traditional forms of plagiarism or ghostwriting (). Emerging scholarship suggests that AI is prompting a transition in academic integrity governance, from detecting misconduct to clarifying normative boundaries, reconstructing standards, and strengthening students’ judgment capabilities (; ; ). In this context, the central issue of academic integrity is no longer whether students endorse integrity norms, but whether they can make defensible ethical decisions in concrete AI use scenarios (). Questions concerning citation, verification, and disclosure in AI-assisted academic work require students to engage in real-time ethical trade-offs under conditions of ambiguity (). Compared with traditional misconduct, AI-related controversies are more closely associated with contextual use, transparency, responsibility allocation, and evidentiary credibility (). Accordingly, the quality of students’ decision-making in complex situations becomes a more appropriate focal outcome. Although prior research has examined AI and academic integrity from perspectives such as governance responses, institutional policies, and student and faculty attitudes (; ; ), explanations of how students form high-quality ethical decisions in specific AI-related contexts remain limited. At the student level, empirical studies have predominantly relied on attitudes, perceived risks, usage intentions, behavioral tendencies, or self-reported compliance judgments to characterize students’ engagement with AI (; ). While these constructs help explain willingness to use AI or concern about misconduct risks, they are insufficient to capture variations in decision quality under ambiguous task conditions. Therefore, compared with general attitudes or behavioral intentions, AI ethical decision-making more directly reflects situational judgment quality and serves as a more suitable core variable for explaining academic integrity differences in the AI era (; ). In this study, AI ethical decision-making refers to students’ ability and performance in identifying normative boundaries, weighing responsibilities and consequences, and making defensible choices in AI-related academic integrity scenarios. This construct differs from moral cognition. Moral cognition refers to students’ understanding, endorsement, and internalization of academic integrity principles and AI-use-related ethical norms, whereas AI ethical decision-making refers to the observable quality of students’ choices and justifications in specific scenarios. Accordingly, moral cognition is conceptualized as the internal psychological mechanism, and AI ethical decision-making is conceptualized as the decision-performance outcome. Given the task-dependent nature of AI use and the susceptibility of self-report scales to social desirability bias, this study adopts scenario-based tasks and performance-based scoring to assess decision quality in realistic or quasi-realistic contexts.