Structural Responsibility in AI Systems: The Creator-Creation Asymmetry as Epistemic Foundation of Human Accountability

Abstract Contemporary debates on responsibility in artificial intelligence systems often treat human accountability as a provisional condition, expected to diminish once AI reaches sufficiently advanced autonomy. This paper challenges that assumption by arguing that human responsibility is not merely temporary or contingent on computational capability, but structurally grounded. Drawing on the MIARO framework (Model of Origin Self-Referential Inference), the paper develops the concept of Creator–Creation Asymmetry, according to which artificial systems, regardless of their operational sophistication, remain dependent on human-designed epistemic and infrastructural conditions. Since AI systems do not possess genuine agency in the full normative sense—including responsibility, susceptibility to consequences, and authentic moral accountability—the epistemic chain of responsibility remains centered on human agents who design, deploy, supervise, and maintain these systems. The paper further argues that even a hypothetical technological singularity would not, by itself, constitute sufficient grounds for transferring structural responsibility from humans to artificial systems.