This study develops and empirically tests a conditional process model to explore the impact of AI recommendations on consumer shopping satisfaction on e-commerce platforms. In addition, this study investigated the moderating effect of consumer value preferences on the relationship between AI recommendations and purchase satisfaction. The findings indicate that AI recommendations cause information cocoons and have a significant negative main effect. However, this negative effect is not uniform. Its impact is significantly moderated by customer value preferences. We further uncover a significant gender divide in these moderating pathways. This study offers actionable insights for e-commerce platforms on how to tailor their recommendation algorithms not just to user profiles, but to user shopping motives.