Key points are not available for this paper at this time.
This paper examines how Artificial Intelligence in Education (AIED) is reshaping teaching and learning, drawing on a systematic literature review alongside policy analysis to explore practical applications, the theories behind them, and their governance consequences. Adopting the EU AI Act’s risk-based lens, we investigate the ways in which regulatory demands—ranging from transparency and data stewardship to human oversight and provider accountability—influence how AIED tools are built and taken up in practice. We group current uses into four areas—adaptive learning, intelligent assessment, learner profiling, and emerging tools—and read them through the prism of well-known learning theories such as constructivism. The analysis underscores that while these technologies hold real promise, several prominent use cases—automated grading and learner profiling, for instance— fall squarely within the EU AI Act’s higher-obligation categories, which means equity, explainability, and genuine human control are not optional but essential for public trust. On the basis of these findings, we put forward concrete, compliance-oriented recommendations aimed at helping educators, institutions, and policymakers deploy AI responsibly across varied educational settings.
Dimić et al. (Wed,) studied this question.