Publication Details
Abstract
The integration of artificial intelligence into digital education has rapidly accelerated, yet the field lacks a cohesive understanding of the methodologies, theoretical frameworks, and empirical evidence guiding this transformation. This systematic review examines artificial intelligence methodologies in digital education through an IMRAD (Introduction, Methods, Results, and Discussion) structure. The study analyzes recent peer-reviewed literature, frameworks, and empirical studies published between 2023 and 2025. The results identify three primary categories of AI methodologies: theoretical pedagogical frameworks for AI integration (including i-TPACK, SAMR, and 3wAI), technical implementation architectures (such as bidirectional personalization systems and GNN-RL hybrid models), and AI literacy development approaches. Meta-analytic evidence demonstrates a significant positive effect of AI technologies on learning outcomes (Hedges' g = 0.86), with generative AI and chatbots showing the strongest effects (g = 1.02). However, significant heterogeneity across studies indicates that contextual factors and implementation strategies critically moderate efficacy. The discussion synthesizes these findings into a integrated model of AI educational methodologies and identifies implications for researchers, educators, and policymakers. This review concludes that effective AI integration requires the deliberate alignment of pedagogical frameworks, technical architectures, and literacy development within specific educational contexts.