IMPACT OF AI-BASED ADAPTIVE LEARNING SYSTEMS ON STUDENT ENGAGEMENT AND LEARNING OUTCOMES

Abstract

Adaptive learning systems that use artificial intelligence have become more popular in contemporary education, but their actual effectiveness in student engagement and achievement in STEM classrooms is not well understood. The paper is the evaluation of an adaptive learning model based on reinforcement learning implemented in a STEM course of a university in eight weeks. We examined the engagement measures among 120 students, the pre/post academic performance, and qualitative interviews of the students and instructors using a mixed-methods design. Findings indicate that adaptive AI system enhanced overall engagement by 28.4%, quiz accuracy by 17.2% and time-to-mastery by 22.9% relative to traditional fixed-path learning. The students indicated increased motivation and satisfaction with the individualized learning paths. These results indicate that adaptive AI systems can greatly increase STEM learning in case they are implemented into the current LMS systems.