The Algorithmic Turn in K-12 Online and Blended Learning: Is AI a Pedagogical Mirage or a New Foundation for Student Agency?

The boundary between online and blended learning in K–12 education is increasingly shaped by algorithmic mediation, raising a central tension: whether artificial intelligence supports student agency or reproduces standardization through new forms of governance. This study reports a computationally assisted critical literature review (CLR) of 31 peer-reviewed articles on artificial intelligence in K–12 online and blended learning. We combine systematic review protocols with text-mining techniques to map dominant intellectual clusters and their conceptual linkages. Specifically, t-distributed stochastic neighbor embedding (t-SNE) is used to visualize proximity-based thematic islands, while lexically connected structures are used to trace relational pathways among core concepts. The analysis identifies five themes organizing the field: predictive governance and the institutionalization of at-risk learners; platformized personalization and agent mediation redefining the blend; automation of assessment and feedback and the efficiency trap; student monitoring and natural language processing as infrastructural surveillance; and acceptance, affect, and trust as emotional conditions of agency. Across themes, the literature emphasizes scalable infrastructures (prediction, platforms, automation, monitoring) more consistently than explicit theorization and measurement of student agency. This paper discusses how these trajectories shift pedagogical authority toward default system logics and provides a synthesized framework of practical translations and governance safeguards to protect learner autonomy in K–12 online and blended settings.