Low learning motivation and limited conceptual understanding of multiplication and division remain persistent challenges in elementary mathematics education, highlighting the need for innovative and engaging instructional approaches. This study examined the effect of Canva AI-based Mathematics Tug-of-War Media integrated with the Deep Learning approach on elementary students' learning motivation and mathematics achievement. The study employed a quasi-experimental pretest–posttest nonequivalent control group design because intact classes could not be randomly assigned. The population comprised all fifth-grade students of SDN 1 Nglanjuk and SDN Sumberpitu. Using total sampling, all 50 students participated in the study: 26 in the experimental group and 24 in the control group. Data were collected using a mathematics achievement test and a learning motivation questionnaire and analyzed using descriptive statistics, N-Gain, ANOVA, and MANOVA. The experimental group achieved a higher mean N-Gain (0.61 ± 0.08) than the control group (0.35 ± 0.09). MANOVA revealed a significant multivariate effect with a very large effect size (Pillai's Trace = 0.810, F(2,47) = 100.257, p < .001, Partial Eta Squared = 0.810). This study contributes to mathematics education by introducing an instructional model that integrates AI-assisted Canva media, a traditional Tug-of-War educational game, and a deep learning approach to improve elementary students' learning motivation and mathematics achievement simultaneously. Future research should examine the long-term effectiveness of this instructional model across different mathematics topics, educational levels, and more diverse educational settings.
Copyrights © 2026