Abstract: Video-based learning (VBL) is a cornerstone of modern education. With the rise of artificial intelligence (AI), tools such as AI-generated video summaries are increasingly used to enhance learner efficiency and streamline content delivery. However, little is known about how such summaries influence learner behavior, cognitive engagement, and motivation during exposure to instructional materials. The present study examined the impact of AI-generated summaries on engagement, comprehension, and intrinsic motivation in a controlled VBL environment. Sixty participants were randomly assigned to either a control group (video only) or a summary group (video plus AIgenerated summary). Results showed that summary access led to significantly reduced video viewing time, suggesting that learners treated the summary as a substitute rather than a complement. While comprehension scores were only moderately lower in the summary group, deeper analysis revealed underperformance on conceptual transfer items. Intrinsic motivation was also significantly lower, particularly in interest and perceived value. These findings underscore the need for intentional design when integrating AI-generated support, as unstructured summary access may promote shallow engagement and diminish learning outcomes. The study concludes with design implications, recommended strategies for integrating AI-generated supports, and directions for future research in VBL environments.