Policies for the use of generative AI in higher education and performance metrics

Abstract

The article considers the formation and assessment of policies governing the use of generative artificial intelligence in higher education, emphasising the relevance of this issue amid the rapid spread of artificial intelligence tools that affect educational processes, academic communication, and assessment practices. The purpose of the study is to systematise conceptual approaches to regulating the use of generative artificial intelligence in higher education institutions and to develop a system of indicators for assessing the effectiveness of its implementation. The methods used are: analysis of the scientific literature to review the current state of the research problem, and generalisation and systematisation to present the study's results. Results. It was established that higher education institutions use three main models for regulating the use of generative AI: permissive, restrictive, and mixed, of which the latter is the most adaptable across different educational contexts. It is considered how ethical principles, such as transparency, identification, and responsibility, shape institutional recommendations and determine acceptable forms of AI use. It is determined that leading universities in the USA, EU, and Asia implement differentiated strategies for the use of artificial intelligence, combining teaching support, mandatory disclosure of AI involvement, and the integration of AI literacy into educational programs. It is noted that practical evaluation of policies for the use of generative artificial intelligence requires a system of indicators grouped by pedagogical, organisational and ethical dimensions, which allows institutions to track both learning outcomes and trends in academic integrity. Conclusions. The implementation of generative artificial intelligence should be accompanied by continuous monitoring of the process, with the participation of all integration subjects, and by regular review of institutional regulations. It was found that the optimised strategy not only prevents academic misconduct but also enhances reflective and autonomous learning. The study notes that the development of comprehensive indicators for the effectiveness of generative artificial intelligence implementation can help inform strategic decisions and improve the quality of education.