Analysis of the ethical aspects of the introduction of artificial intelligence into the educational process of technical universities and measures to minimize risks

Abstract

The study aims to identify and systematize ethical risks arising from the implementation of artificial intelligence technologies in the educational process of technical universities, as well as to substantiate measures for their minimization with regard to academic values and institutional constraints. The empirical and theoretical foundation includes the analysis of regulatory documents, codes, and digital transformation strategies of leading universities, a review of academic literature on AI ethics and the philosophy of education, a comparison of typical AI application practices (adaptive learning, algorithmic assessment, proctoring, learning analytics), and discourse analysis of public narratives about the neutrality of algorithms. The research applies methods of conceptual reconstruction of key categories (justice, autonomy, responsibility), stakeholder mapping, and matrix-based risk assessment using scales of probability, harm, and controllability. The results demonstrate that the main threats are associated with algorithmic bias and the reproduction of social inequality under the guise of personalization; the erosion of students' and teachers' epistemological and pedagogical autonomy due to the platform logic of nudging and total monitoring; the diffusion of responsibility in human-algorithm systems, creating a vacuum of accountability; as well as the reduction of educational goals to efficiency metrics, which leads to the dehumanization of learning. A multi-level set of measures is proposed: institutional (permanent AI ethics committees, procedures for preliminary and subsequent humanitarian expertise, transparent data policies, and mechanisms for appealing algorithmic decisions), pedagogical (integration of modules on digital ethics and data criticism into curricula, development of teachers' competencies in interpreting algorithmic recommendations, preservation of human zones of assessment), technological (dataset auditing, model documentation, assessment of impacts on vulnerable groups, the privacy by design principle), as well as mechanisms of distributed responsibility with clearly defined roles for developers, administration, teachers, and students. The conclusion emphasizes that ethically sound AI integration is possible when humanistic goals of education are prioritized, accountability is institutionalized, and continuous interdisciplinary reflection is maintained, ensuring a balance between innovation and the protection of the rights of participants in the educational process.