Abstract
Integrating AI into educational systems is often approached through either pedagogical or technological lenses, leaving governance and contextual dimensions insufficiently addressed. In this study, we address this need by proposing a four- stage cyclical framework for AI integration in education—assessment, design, implementation, and evaluation—that builds on a constructivist learning theory and socio-technical systems approach and incorporates a system of ongoing feedback and evaluation in the AI integration process to ensure that ethical considerations are taken into account at every stage. The proposed framework is informed by a narrative literature review of studies on AI integration in education from pedagogical, technological, and governance perspectives, based on a search of the literature databases Web of Science, Scopus, and ERIC. While our discussion is focused primarily on the Turkish context, where issues such as examination-centred education, infrastructure and digital literacy inequalities, legal and institutional barriers, and teacher digital literacy are obstacles to AI integration in education that have received relatively little attention in the literature on global AI integration frameworks, our proposed framework is designed as a context-sensitive model that may be adapted to different educational contexts, although its applicability across diverse systems requires further empirical examination. Empirical research using this framework to develop, implement, and evaluate AI-enhanced educational environments in a range of educational settings and cultural contexts will be needed to further test and refine the model proposed here.
Keywords: AI in Education, Edtech, Pedagogy, Turkish Education System.