Generative artificial intelligence (AI) is rapidly transforming higher education, influencing how students learn, produce knowledge, and engage in assessments. This study examines the ethical implications of AI integration in university courses and its perceived impact on students’ cognitive skills. Using a mixed-methods secondary data analysis, survey data from 23,218 students across 109 countries and 252 teachers from China, Romania, India, and Brunei were analyzed to explore perceptions of AI capabilities, ethical boundaries, behavioral intentions, and cognitive outcomes. The results have shown that although students are aware of the potential of AI in improving the efficiency and content creation and assistance in the learning process, there are still concerns about overdependence, plagiarism, and lack of critical thinking. The views of faculty underline moral direction and the necessity of systematic assimilation. Developing upon these insights, a theoretical framework is developed that connects antecedent variables, behavior intentions and institutional context to two outcomes the cognitive impact and institutional influence. This framework offers a guide on how universities should introduce generative AI in a responsible manner, and strike a balance between ethical principles and training higher-order cognitive abilities.
This methodological and conceptual contribution develops an integrative framework for research on educational technology that systematically links internal and external validity, different research approaches, and heterogeneous outcome dimensions. Drawing on the recurring disappointment regarding the limited learning benefits of digital technologies— illustrated by the example of virtual reality (VR)—the article argues that technological innovations can only be adequately assessed if their effects can be clearly attributed and transferred to authentic learning contexts. Using central research logics, including media comparison, value-added, and ATI/LTI approaches, it shows how the choice of research approach shapes the type of evidence generated and the associated validity requirements. In addition, the article extends existing outcome models by incorporating a behavioral dimension to reflect the multidimensional nature of teaching and learning processes. The proposed framework illustrates that progress in educational technology research depends on the combination of methodological rigor, context-sensitive evidence, and multidimensional outcome assessment. Finally, practice-oriented recommendations are presented to support cumulative evidence development, transparent reporting standards, and the integration of laboratory and field research.
In the original version of this article (https://doi.org/10.3991/ijet.v18i12.39697), the affiliation for the first author, Jianmeng Ou, was incorrectly published as “Hainan University, Haikou, China”. Following a request from the author, this has been corrected to the institution where the research was conducted: “Tianjin University, Tianjin, China”. On 2026-01-08, the original article has been updated to reflect this change.
In the original version of this article (https://doi.org/10.3991/ijet.v18i14.40397), the affiliation for the first author, Hongbin Lin, was incorrectly published as “Hainan University, Haikou, China”. Following a request from the author, this has been corrected to the institution where the research was conducted: “Tianjin University, Tianjin, China”. On 2026-01-08, the original article has been updated to reflect this change.
Virtual reality (VR) has been proposed as a promising tool to enhance social competences in higher education, especially in resource-constrained counseling seminars. This study investigates the effects of a single-session VR-based training on university students’ counseling competence and self-efficacy. Forty-five students took part in a standardized counseling scenario delivered through a realistic social VR environment, interacting with a trained actor in real time. Pre- and post-training questionnaires assessed changes in self-efficacy and counseling competence, alongside measures of presence, realism, cognitive load, and perspective-taking. Results showed small, non-significant improvements in both outcome variables. Regression analyses revealed that perceived realism significantly predicted counseling competence, while presence and extraneous cognitive load influenced perspective-taking. However, the limited duration of the intervention and the absence of a control group constrain causal inferences. Despite these limitations, the study provides initial empirical support for the integration of VR technologies into counselor education and outlines key factors for effective implementation. Future research should explore adaptive, Artificial Intelligence (AI)-enhanced VR environments that allow for repeated practice and personalized feedback to fully realize the potential of VR in social skills training.
