Integrating Explainable Artificial Intelligence Into Adaptive Personalized Learning Method

Abstract

As artificial intelligence (AI) becomes deeply integrated into educational technologies, adaptive learning systems offer scalable, personalized instruction. However, the opaque nature of many AI-driven platforms hinders learners’ ability to understand and trust system recommendations, an issue that is especially pressing in under-resourced secondary schools, where algorithmic opacity can worsen existing inequities. This study examines the effects of an adaptive learning system embedded with XAI features on learner performance, engagement, and perceived agency. A mixed-methods quasi-experimental design was used to compare the outcomes of 350 secondary school students using a traditional non-adaptive platform and an AI-powered adaptive system with XAI. The quantitative data included pre- and post-test scores, engagement logs, and motivation and trust scales. Qualitative data were collected through interviews and think-aloud protocols and analyzed thematically using Braun and Clarke’s framework. The results revealed that students using the XAI-enhanced system showed significantly higher learning gains (p<.05), improved engagement, and a clearer understanding of their learning trajectories. Four themes emerged: enhanced trust via algorithmic transparency, alignment of AI feedback with personal goals, usability barriers, and cultivating reflective learning habits. Notably, students in resource-limited settings responded positively to system explanations, highlighting XAI’s potential of XAI for equitable digital learning. The integration of XAI not only boosts academic outcomes but also nurtures learner trust, autonomy, and motivation. Ethical considerations, such as fairness, cognitive load, and cultural adaptability, are also highlighted, underscoring the importance of human-centered design. These findings advocate for a human-centered, transparent design in educational AI, which is critical for inclusive adoption in low-resource environments.