Adaptive Difficulty and Stealth Assessment in Collaborative Game‐Based Learning

ABSTRACT
This article explores the application of reinforcement learning‐based dynamic difficulty adjustment (DDA) algorithms in collaborative game‐based learning environments, with a focus on intelligent assessment. Adaptation in gaming environments is essential for providing personalized learning experiences that adapt to a wide range of learner needs. Although DDA algorithms are commonly used to adjust game difficulty for individual performance, research on their effectiveness in collaborative settings remains limited. Our study addresses this gap by proposing a novel reinforcement learning‐based DDA algorithm that integrates real‐time performance data from both individual and group interactions, enabling dynamic adjustments that maintain an optimal balance between learner challenges and skills. Additionally, we introduce the GRADES framework, a layered architecture that combines adaptive decision‐making, stealth assessment, and continuous performance monitoring to personalize learning experiences at both individual and group levels. Comprehensive simulations and comparative analysis of existing DDA algorithms show that our approach improves engagement and learning results across a range of game difficulty levels. These findings highlight the possibility of integrating reinforcement learning and stealth assessment to develop adaptable, responsive educational environments, thereby advancing the field of collaborative game‐based learning.