Navigating the Future of Learning: A Systematic Review of AI-Driven Intelligent Tutoring Systems (ITS) in K-12 Education

Abstract The use of artificial intelligence in education (AIEd) has grown exponentially in the last decade, particularly using intelligent tutoring systems (ITSs). Despite the increased use of ITSs and their promise to improve learning, their real educational value remains unclear. Therefore, this systematic review aims to identify which experimental designs are currently used to evaluate the effects of ITSs and what are the effects of ITSs on K-12 students' learning and performance. Studies retrieved from ERIC USDE and Scopus databases reporting effects of an ITS in a formal learning setting published after 2009 were screened and the data extracted was categorized by experimental design. The 20 studies analyzed in this systematic review included a total of 2853 students (N = 2853), predominantly in middle and high school Science, Technology, Engineering and Mathematics (STEM) -related classes. These studies used quasi-experimental designs with varying intervention durations, with half of them categorized as very short. Some studies reported positive effects of ITSs on learning and performance in contrast to traditional teaching. However, the effects were found to be mitigated when compared to non-intelligent tutoring systems. The included studies have identified optimal conditions or contexts for the application of specific ITSs. Overall, our findings suggest that the effects of ITSs on learning and performance in K-12 education are generally positive. However, additional research with longer interventions and increased sample sizes with greater diversity is warranted. Additionally, the ethical implications of using AI for teaching should be investigated.