Adaptive learning environments employ artificial intelligence (AI) to personalize education by dynamically adjusting task complexity in response to learner performance. While prior research emphasizes the motivational benefits of these systems, the interaction between flow and cognitive load remains insufficiently examined. Drawing on Flow Theory and Cognitive Load Theory, this study investigates how AI-driven task complexity adaptation shapes engagement, cognitive effort, and learning performance. We hypothesize that cognitive load, when maintained within an optimal range, fosters flow experiences that enhance learning outcomes. A between-subjects experimental design with 560 business school students compares adaptive and static learning conditions using OATutor, an AI-driven platform. Cognitive load is assessed through validated self-report measures, while flow is measured through both electroencephalography (EEG) and psychometric scales. The findings will advance theoretical understanding of adaptive learning and inform the design of intelligent systems that balance engagement and cognitive efficiency for improved learning performance.