Abstract:

This systematic literature review examines how contemporary Artificial Intelligence driven learning systems implement adaptive and personalized educational content, with particular attention to the role of intelligent agents, learner modelling and system level adaptation mechanisms. Focusing on peer reviewed studies published in the last six years, the review identifies and analyse research that moves beyond static personalization toward dynamic, data driven forms of adaptation. Rather than treating personalization as a generic design goal, the study investigates how adaptation is operationalized in practice, including what is adapted, when adaptation occurs and which learner data trigger adaptive decisions during learning interaction. The review further analyses how learners are represented within these systems, examining the types of learner variables modelled, the structure and dimensionality of learner models, the data sources used and the degree to which learner models are operationalized in functioning systems. The findings indicate a clear shift toward continuous and longitudinal adaptation driven mainly by performance and interaction data, alongside a strong prevalence of fully automated adaptation. At the same time, important gaps are identified, particularly in the modelling of affective and metacognitive learner characteristics, the involvement of educators during runtime adaptation and the transparency of adaptive decision making. Overall, the review provides a structured synthesis of current design patterns and limitations in AI driven adaptive learning systems, offering conceptual clarification and empirical insight to inform future research, system design and evaluation.

Keywords:

Adaptive learning, Personalized education, Artificial Intelligence in education, Intelligent agents, Low-code platforms, Workflow automation, Learner modelling, Large Language Models (LLMs), Explainable AI, Educational technology.