Unintentionally Breaking the Fifth Wall: How Generative AI Invites Metalepsis in Real Life

This position paper explores how the use of generative AI in storytelling introduces unwanted instances of metalepsis (disruptions of narrative boundaries). Through the prism of "fifth wall breaks", we examine a variety of examples of how narrative AI systems unintentionally generate inconsistencies or self-references, shattering users’ referential illusion and sense of immersion. The paper also introduces the concepts of "real-life metalepsis", illustrated by additional examples where users abruptly encounter the artificial nature of AI interactions, breaking the “relational illusion” and potentially causing confusion. Future research will investigate how to mitigate these disruptions and explore the creative integration of AI-induced metalepsis into narrative design.