Bridging Pedagogy and Computation: A Theoretical Framework for AI in Educational Systems

Abstract

Introduction

Today the integration of Artificial Intelligence (AI) in education systems is a paradigm shift in in building, delivering, and assessing knowledge. Though many studies address AI applications and implications, their theoretical underpinnings are scattered across disciplines. AI-powered adaptive learning software, intelligent tutoring systems, automated grading, and predictive analytics are being used in educational institutions increasingly, generally without an explanatory theoretical understanding of how these technologies alter learning processes.

Objectives:

This paper synthesizes key theoretical perspectives of constructivism, cognitive load theory, personalized learning, and socio-technical systems theory, to build an integrative framework explaining AI’s pedagogical role. It also addresses how AI-backed systems operate and confront learning theories, focusing on the interplay among technological affordances, learner agency, and institutional dynamics.

Methods:

An extensive review of the literature was conducted across learning sciences, cognitive psychology, information systems, and critical theory, including articles published between 2013 and 2025. Adaptive learning systems, intelligent tutoring systems, and AI-based personalisation models are discussed to reveal cognitive processes and implementation implications at individual, social, and institutional levels.

Results:

Five theoretical frameworks were recognized as fundamental: constructivism for AI-enhanced learning environments, cognitive load theory for the design of intelligent systems, connectivism and networked intelligence, adaptive personalization theory, and socio-technical systems theory. Main results suggest that AI can be a cognitive tool, instructional mediator, social agent, and institutional technology, yet a multi-level theoretical appreciation of its educational use is required.

Conclusions:

AI learning needs to be envisioned through a composite theory model that combines cognitive, social, and institutional aspects. The model suggested here focuses on personalized scaffolding, transparent and explainable algorithms, equitable and bias-aware design, and alignment between technological efficacy and genuine learning objectives. The study finally underscores the imperatives of theory-driven innovation where pedagogical theory would inform AI innovation for improving learning without undermining human agency and equity.