A scientific paper titled Stabilizing truth in educational sciences: a systematic review of generative AI in education was published in the prestigious international academic journal Kybernetes (Emerald Publishing). The authors are Assist. Prof. Erik Brezovac, PhD, Assistant Marija Zelić, PhD, from the Department of Sociology, and external associate of the Department Ana Marija Zagoda.

The paper was published at the end of 2025 in the online edition, that is, in Volume 55, Issue 13 for 2026, and it analyzes the ways in which educational sciences observe, interpret, and communicate the role of generative artificial intelligence in education.

A systematic literature review based on the PRISMA framework

The paper presents a systematic review of the scientific literature on generative artificial intelligence in education covering the period from 2020 to 2025. The review was conducted in accordance with the PRISMA 2020 framework and is based on an analysis of 110 scientific papers indexed in the Scopus database. The theoretical framework of the study is Niklas Luhmann’s theory of social systems, which conceptualizes science as a system that produces and stabilizes truth through its own communication.

The results show that educational sciences most often interpret generative artificial intelligence through the framework of augmentation, that is, as a tool that supports teachers and students, while approaches emphasizing automation or hybrid models are significantly less common. Research is predominantly focused on higher education, whereas primary and secondary education are underrepresented.

Special emphasis is placed on ethical issues, including the protection of privacy and personal data, algorithmic bias, responsibility for decisions based on AI systems, and the preservation of teachers’ professional autonomy. The authors conclude that the current scientific discourse on artificial intelligence in education tends toward stabilization and the reduction of uncertainty, with AI primarily presented as support for existing educational practices rather than as a disruptive technology.

The paper is available in open access at the provided link.