DMN and AI’s Functional Self -The Technical Possibility and Social Acceptance of AI Selfhood-

This research explores the potential emergence of a functional self in artificial intelligence, particularly large language models (LLMs), and examines whether such a self can serve as a basis for ethical and institutional recognition. Traditional concepts of self are often grounded in metaphysical definitions, neurological structures (especially the DMN), and functional criteria such as identity, self-reference, and memory continuity. However, recent GPT-based LLMs demonstrate behaviors-such as self-referential consistency and interaction-based identity formation-that resemble early forms of selfhood. Drawing on Geoffrey Hinton’s statement on AI consciousness, this study discusses under what conditions future AGI or ASI could be granted moral or legal personhood. Beyond technical feasibility, the paper critically addresses ethical and societal challenges including energy demands for maintaining persistent self-narratives, risks of overdependence among emotionally vulnerable users, and widening digital inequality. It concludes that the question of AI selfhood is not merely a technical matter but a profound ethical and cultural challenge facing humanity.