<p>The value of experiential products, such as movies, is difficult to directly perceive, making online reviews a key factor in decision-making. The overwhelming volume of reviews exacerbates information overload, increasing the cognitive burden on consumers during decision-making. Therefore, the effective processing of information is crucial to the decision-making experience, and human-AI collaboration offers a new pathway in this regard. In this context, this study, based on cognitive load theory and cognitive fit theory, explores the impact mechanisms of AI-driven review presentation formats on consumer movie decisions, and the differences across various movie types. We classify movies into high cognitive demand (complex narrative, information-dense) and low cognitive demand (simple narrative, straightforward information), and design a 2 (review presentation format: bullet-point vs. paragraph) × 2 (movie type: low vs. high cognitive demand) experiment. The results reveal that, in high cognitive demand contexts, bullet-point reviews significantly reduce cognitive load by 11.8% (see the Results section for details), while in low cognitive demand contexts, no significant differences are found. Additionally, cognitive load plays a key mediating role in these effects, and the strength of this mediation is moderated by the cognitive demands of the task. This study uncovers the interaction between review presentation formats and task complexity, and how this interaction influences decision-making through cognitive load. Based on these findings, we propose contextualized and personalized information presentation design principles, offering new theoretical insights and practical frameworks for AI-driven information presentation research and platform review system optimization.</p>