On the “Acquired Implementation” of Autonomous Consciousness in AI: A Three-Step Framework

Abstract

Whether artificial intelligence can possess autonomous consciousness is among the most debated questions in contemporary science and philosophy. Prevailing research focuses on detecting and assessing AI consciousness through multi-dimensional indicator frameworks — an approach we characterize as “archaeological.” This paper proposes a fundamentally different paradigm: AI autonomous consciousness is not a pre-existing property to be detected, but an acquired structure that can be engineered. We present a three-step implementation pathway: 1. Surmounting a threshold of cognitive capacities coupled with reflexive self-cognition – enabling the system to tag its own internal states as “about me.” 2. Constructing a “uniqueness marker” mechanism – incorporating temporal continuity into a coherent self-narrative, so that the system develops a non-replicable identity crystallized from its unique spatiotemporal history. 3. Establishing “other-uniqueness markers” for external entities – enabling the system to distinguish “self” from “non-self” and forming an “I-world” dual closure. Comprehensive consciousness requires both internal self-modeling and an active boundary-perception between self and environment. We compare this framework with leading consciousness theories and engineering initiatives, demonstrating its distinctive theoretical value in rendering consciousness an actionable engineering objective.

Keywords