As generative AI becomes the primary interface to information, recommendation systems are quietly transforming into engines of interpretation. Instead of simply filtering content, GenAI systems increasingly define how knowledge is framed, explained, and contextualized—creating personalized realities at scale.
This talk explores the emergence of a new kind of filter bubble, one rooted not in what is shown, but in how meaning is constructed. It examines the societal, cultural, and economic implications of AI systems that adapt narratives to individual users, potentially narrowing perspectives while appearing more helpful than ever.
By reframing recommendation systems as cognitive infrastructure, the session argues for a shift in design philosophy—from maximizing engagement to preserving pluralism, serendipity, and trust. The talk concludes by outlining how industry leaders can ensure that GenAI-powered recommendations expand human understanding rather than constrain it.
