Introduction:

Abstract Artificial intelligence in education is widely promoted as a solution for personalizing learning and advancing equity. However, this paper argues that AI-driven adaptive learning systems systematically embed and amplify existing socioeconomic disparities through algorithmic bias, data colonization, and the commodification of learning pathways. While these systems generate the appearance of individualization through dynamic content delivery and responsive feedback, their underlying logic reinforces rather than disrupts educational hierarchies. Trained on datasets reflecting decades of educational stratification, these algorithms learn to recognize and predict patterns rooted in prior inequity. This research employs critical algorithmic analysis and comparative case studies of three major adaptive learning platforms to examine how personalization mechanisms—algorithmic classification, predictive modeling, and pathway recommendation—encode and reproduce structural inequalities at scale. Findings reveal that students from privileged backgrounds receive sophisticated, exploratory personalization promoting higher-order thinking, while marginalized students are funneled into rigid, remedial trajectories based on historical performance data. Rather than democratizing education, these systems create a tiered educational landscape that naturalizes inequality through technological legitimacy. The paper concludes that the paradox of personalization reflects not technical limitations but fundamental design choices reflecting corporate interests and neoliberal educational logics. Addressing this requires moving beyond algorithmic transparency toward structural interventions that question whether personalization through AI can ever serve equity without radical reimagining of educational technology’s political economy. ...