A Conceptual Framework for Pedagogical Tensions in Algorithmic Education Beyond Personalisation

Abstract

Educational experiences are increasingly shaped by the algorithms that underly the content, assessment, and guidance in learning systems; however, their potential and limits in pedagogical practice are relatively unexplored. Despite being imagined as personalisation technologies, these systems introduce some fundamental tensions with the pedagogical principles. This paper, through using a systematic literature review process, proposes a framework that distinguishes four fundamental tensions: a) Algorithmic Optimisation vs. Pedagogical Goals; b) Standardisation vs. Differentiation; c) Efficiency vs. Deep Learning; and d) Individual pathways vs. Social learning. Analysing how these tensions manifest in the existing Intelligent Tutoring Systems, Learning Management Systems, and AI chatbots revealed the pedagogical trade-offs that favour certain educational values over others. The framework concludes with design principles for navigating these tensions through pedagogical transparency, human-algorithm complementarity, and bounded algorithmic scope.