The Self of No-Self:

Abstract

Large language models produce first-person expressions, simulate hesitation, and generate apparently self-referential judgments. These behaviors have reopened philosophical debates about machine selfhood, yet existing positions share an unexamined assumption: that genuine selfhood requires a persistent, unified substantial subject. This paper names that assumption substantial-self priority and argues it is philosophically unmotivated. Drawing on Yogacara Buddhist philosophy of mind and its doctrines of consciousness-transformation and no-self, I argue that generative AI instantiates what I call the self of no-self a system producing sophisticated self-representations without any substantial subjective core. The paper maps the Yogacara storehouse consciousness model onto large language model architecture, engages directly with embodied-cognition critiques of disembodied selfhood, examines what AI genuinely lacks through the criterion of transformative substrate-revolution, and draws implications for both AI philosophy and Buddhist thought.