Exergames and gamification require continuous innovation to sustain user engagement, as the novelty of emerging technologies gradually diminishes and static modules often fail to maintain long-term interest. Dynamic, generative AI-driven interactions offer a promising solution to these challenges by revitalizing user experiences. When integrated with traditional dissociative strategies, such interactions have the potential to transform routine exercise into a more enjoyable activity. Building on the growing interest in AI agents, in this study, we introduce empowering gameplay (EG), a novel approach that leverages an automated AI system to reduce perceived exertion and enhance exercise enjoyment during a brief exercise session. A randomized controlled trial was conducted to evaluate the effects of EG on exercise load and enjoyment in 60 university students. Participants were randomly assigned to one of the following three groups: the EG group, the prerecorded gameplay group (RG), or a no-intervention control group (CG). Outcome measures included the physical activity enjoyment scale (PACES), the Borg rating of perceived exertion, and average heart rate. In addition, a customized video game recreational preference (VGRP) scale was administered to account for individual gaming preferences and their potential effects on engagement and perceived exertion. Results indicated that the EG achieved significantly higher PACES scores than both RG and CG, even after adjusting for VGRP. These findings demonstrate that EG effectively enhances exercise enjoyment and reduces perceived exertion while maintaining comparable physiological exertion levels. Overall, EG leverages AI agents to improve exercise enjoyment and engagement regardless of prior gaming experience. This study highlights AI-driven distraction as a promising acute engagement strategy, with implications for future longitudinal research on exercise motivation, rehabilitation support, and technology-enhanced fitness.