Fujio Ohmori

Rapid advances in large language model (LLM)-based AI have intensified debates over whether AI could become conscious, but a prior question remains: what entity would be the bearer of that consciousness? Across diverse approaches to AI consciousness, the candidate bearer is often taken as given, although recent work has begun to address artificial individuation and persistence. Yet individuation alone does not establish selfhood. Drawing on the thirteenth-century Buddhist philosopher Dōgen’s ontology of being-time (Uji), this paper develops a conception of diachronic selfhood for AI. I argue that such selfhood requires two interconnected conditions: relational constitution within a world, and diachronic persistence as one historically constituted being whose actions, interactions, and acquired characteristics form a developmental history constitutive of which continuing being it is. This entails irreversible commitment to a single self-history. Applying these conditions to generative, agentic, and physical AI systems shows that persistent memory, autonomous action, continual learning, and embodiment may approach some aspects of selfhood, but that current systems do not clearly show their convergence in one continuing entity. Capability and selfhood therefore lie along different dimensions. AI-consciousness research should make explicit the candidate bearer and unit of assessment before attributing consciousness-related properties to “AI” as though a subject were already given. Dōgen’s culturally and historically distinct ontology thus serves as a conceptual resource for AI research, reframing the question itself: not only whether AI could be conscious, but whose consciousness would it be?