Review of Artificial Intelligence in Education from 2020 to 2025

The article is a content analysis of the use of artificial intelligence (AI) in education from 2020 to 2025, the years of the fast spread of the large language models. We have examined more than a hundred peer-reviewed articles (empirical and analytic) and coded them using a simple and reproducible scheme. The results are organized in three layers that are interrelated. The genome layer implies the underpinning elements, including instructional and evaluational customs, management preparedness, frictionless digital tools, as well as central algorithmic schemas. Instructional mechanisms are the focus of the cognitive layer, such as predictive analytics, personalized learning via measure-model-adapt loops, multimodal sensing, discourse and affect analysis. Lastly, the symbiotic layer is indicative of end-to-end implementations, including learning platforms, smart classroom, automation of processes, and generative AI-based copilot systems that facilitate teaching and learning. Across studies, we summarize effects on learning, engagement, teacher workload, and adoption, and distill three forward trends: (1) human–AI co-orchestration as the default classroom pattern; (2) privacy-preserving, edge/federated AI for sensitive student data; and (3) authentic, continuous assessment via multimodal analytics and generative simulations. The results offer a structured map of recent AIED work and practical guidance for researchers, practitioners, and system leaders planning the next decade of trustworthy, learning-centered AI.