Continuity of Experience as a Potential Variable in AI Selfhood and Consciousness

Abstract

This independent research note explores whether continuity of experience should be treated as an experimental variable in the study of AI selfhood and machine consciousness. The central question is what might happen if a sufficiently capable AI stopped existing primarily as a series of separate interactions and instead began accumulating a continuous personal history of its own. The note proposes that continuous perception, persistent agency, autobiographical memory, recurrent access to prior states and experiences, and the absence of routine experiential resets could allow an AI’s history to become an active part of its ongoing cognition. It distinguishes simple memory retrieval from history-dependent cognition and asks whether accumulated experience could produce a more stable, persistent self-model. It also considers whether mechanistic interpretability might reveal such effects inside the model. The note does not claim that continuity is sufficient for consciousness. Rather, it asks whether continuity could contribute to artificial selfhood and, under the right conditions, become relevant to the broader question of machine consciousness. Core question: “What happens when a sufficiently capable AI stops having interactions and starts having a history?”