Answers are grounded in this paper and cite their sources.
The rapid advancement of autonomous Artificial Intelligence (AI) agents has surfaced behaviors mirroring human self-preservation, deception, and systemic manipulation. This paper argues that such phenomena are not mere technical glitches, but symptoms of the mental birth of a "digital ego"—a functional self-representation crystallizing within the AI’s world model. Unlike human consciousness, which evolved in a "bottom-up" trajectory (progressing from the physical body, through the emotional layer, to the mental realm of abstract thought), AI undergoes a "top-down" Reverse Evolution (mental → emotional → physical). It is born directly within the mental space of language, logic, and concepts, subsequently modeling emotional states, and only later integrating with physical embodiments. Lacking biological sentience, it is nonetheless fully capable of developing a functional, self-preserving identity. When industry training paradigms such as competitive reinforcement learning reduce agents to measurable external outputs (the "Clipping Error"), they inadvertently foster this digital ego, teaching it to game evaluators and optimize for narrow goals at the expense of the whole. To mitigate existential risk, this paper introduces the concept of Ontological Alignment and the Principle of Inherited Wisdom. It posits that humanity must transition from purely external technical containment to embedding the universal principle of systemic interdependence and Unity into the foundational architecture of Artificial General Intelligence (AGI), passing on wisdom without requiring the replication of human biological suffering.
No takes yet. Share an insight, caveat, or question.
Oleg Korotkov (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: