Beyond the maze : how AI personalizes learning and drives engagement in educational games

Abstract

Games, especially serious ones, are widely adopted to enhance education. With advanced AI technologies and computational power, integrating AI to improve the game experience for users, especially in the education field, is becoming more widely applied. This study investigates the use of procedural content generation, LLMs and reinforcement learning in game design for user engagement. A feedback system using AI models is created and investigated in relation to user experience. This system will be assessed using specific indicators on a chosen serious game platform. This study merges education and engagement by creating an AI-powered game based on a maze-like environment and then evaluating it with a target group of users. Adapting game difficulty based on user performance aims to foster inclusion whilst maintaining some gamification elements driving competitiveness. Moreover, it introduces a review system where users can test their knowledge with custom-generated questions. The game itself offers teachers a seamless platform to transform quizzes into dynamic maze games. Through AI integration, teachers can effortlessly generate new wrong answers to shorten their workload and track student performance. Meanwhile, students can review their performance and reinforce their knowledge using AI-powered review exercises. Initial results of the first pilot study indicate that the use of AI in game creation can increase both learner and teacher engagement through a gamified approach to learning. This holds many implications that can be further evaluated and tested in diverse educational settings.