Design of an AI-based system with dynamic difficulty adjustment and real-time feedback for personalized adaptive learning

Abstract

The paper presents the design and experimental evaluation of an adaptive digital learning system based on artificial intelligence, which involves the integration of personalization algorithms, a recommendation module, and a mechanism for dynamically adjusting the complexity of tasks. The proposed system used user behavioral data and digital interaction analytics to compile individualized learning trajectories in real time. The system architecture includes analytics modules, a recommendation mechanism, and an adaptive complexity adjustment algorithm. The experimental evaluation of the system was conducted in four academic groups (180 users) within the framework of a quasi-experimental design. Statistical modeling methods, Learning Analytics data analysis, and multiple regression were used to analyze the effectiveness. The results showed a statistically significant improvement in learning outcomes in groups working in the adaptive AI mode (η² = 0.60). In addition, a moderate group × time interaction effect was recorded (η² = 0.19). Analysis of behavioral metrics indicated that the integration of AI modules made it possible to increase the intensity of user interaction with the system. The main predictors of learning effectiveness were the intensity of interaction with the system's recommendations and the total time of active work in the platform. The personalized recommendations module had the greatest impact on the results. Therefore, the integration of adaptive algorithms and recommendation mechanisms made it possible to determine the importance of digital learning architecture.