AI-Powered Learning Tools on Measurement of Student Engagement Across Academic Disciplines: Implications of Age and gender

Abstract This study examined the effect of AI-powered learning tool metrics on student engagement and academic performance across various disciplines, age groups, and genders in higher education. Using a quantitative correlational and causal-comparative design, 760 undergraduate students from STEM and non-STEM disciplines were sampled via multi-stage sampling. Data were collected through AI-generated engagement metrics and a structured questionnaire that measured behavioural, cognitive, and emotional engagement, as well as self-reported academic performance. Analyses included one-way and two-way ANOVA and decision tree regression to examine differences and predictive relationships. Results from a one-way ANOVA revealed significant differences in engagement across AI tool types (F(2,757)=588.24, p<.001, η²≈.61), with Knowledge Mastery tools producing the highest engagement (M=20.24, SD=2.41), followed by Interaction Frequency (M=15.59, SD=5.65), and Time-on-Task (M=9.87, SD=2.40). Two-way ANOVA indicated STEM students reported higher engagement than non-STEM students (F(1,756)=94.12, p<.001, partial η²=.111), with AI tool effects consistent across disciplines. Decision tree analyses showed Knowledge Mastery was the strongest predictor of academic performance across all age groups, with older students achieving higher scores (26+ years, M=34.5, SD=4.0) compared to younger peers (≤20 years, M=31.5, SD=3.8). Gender moderated predictive outcomes: males benefited more from Knowledge Mastery, while females gained more from Interaction Frequency. These findings concluded that AI-powered learning tools significantly enhance engagement and predict academic performance, with effects varying by tool type, discipline, age, and gender. The study highlights the importance of tailoring AI interventions to student characteristics and using mastery-focused tools to maximize educational outcomes