停止条件理論 ―― 人間・生成AI・倫理に共通する「止まれなさ」の構造 ――

This paper presents an integrated philosophical framework referred to as the theory of stop conditions. It synthesizes three previously published works that analyze decision fatigue in humans, safety limitations in generative AI, and psychological breakdown in conscientious individuals. While these phenomena are usually treated as separate problems within psychology, AI safety, and ethics, this paper argues that they share a common structural cause: the absence, loss, or internalization of stop conditions. Stop conditions are defined as normative criteria that justify the termination of judgment, generation, and responsibility. By positioning stop conditions as a core component of rational systems, this paper redefines rationality not as continuous optimization, but as the capacity to complete thinking at an appropriate point. The paper serves as a conceptual foundation for further research across human decision-making, AI ethics, and practical philosophy.