The rapid advancement of artificial intelligence has created new challenges for mathematics educators seeking assignments that promote authentic student engagement rather than AI-assisted completion. This study examines student-created video solution homework as an AI-resistant assignment in undergraduate mathematics. Implemented across four undergraduate mathematics courses, the assignment required students to explain their problem-solving processes verbally and visually. Classroom-based empirical evidence was collected through two measures: (1) comparisons of exam performance by video homework completion and quality, and (2) a student perception survey. Across courses, students who consistently earned full-credit video scores demonstrated higher average exam performance than peers with incomplete or lower-quality submissions. Survey responses further indicated that students perceived improvements in conceptual understanding, organization of reasoning, and confidence in explaining mathematics. Together, these findings suggest that structured video explanation assignments can promote genuine engagement and deeper learning while serving as a practical AI-resistant assessment strategy in undergraduate mathematics.

This talk is based on ongoing work, and the manuscript is in preparation.