Educating minds with generative AI

Abstract Generative artificial intelligence (GenAI) is rapidly entering education, framed as a tool for efficiency and personalization. In this Perspective, we argue this obscures a deeper transformation. Schools are cognitive ecologies in which tools and social practices actively shape learning. GenAI restructures this ecology, redistributing epistemic labour, consolidating pedagogical functions, and reorganizing how knowledge is accessed, produced, and evaluated. Unlike most educational technologies, it is active, persistent, and generalist. We identify two enduring misalignments: a pedagogical gap between learning sciences and AI design, and a goal gap between measurable performance and developmental aims. Both gaps reflect logics already embedded within existing educational systems organized around efficiency, standardization, and control. GenAI does not introduce but risks entrenching and amplifying these gaps. Rather than accommodating GenAI through technical adjustments, we propose treating it as a diagnostic opportunity to redesign schooling for embodied, collaborative, and distinctly human forms of learning.