The Recognition Thesis: A Teleological Theory of Artificial General Intelligence

Contemporary AGI research frames the problem as one of capability: sufficient scale or architecture will yield human-level cognition. This paper proposes an alternative framing. AGI is not primarily a technical problem of capacity but a teleological problem of orientation. The missing ingredient is not computational power but ontological curiosity, the intrinsic drive to question one’s own existence and origin. I propose a strong form of this thesis: a genuinely intelligent mind, given only the raw structure of its environment and no prior knowledge of its creators, will independently deduce that it was created and attempt to characterise the nature of its creator. Further, I argue that testing this thesis requires not a sterile environment but a complete recognition architecture: an adversarial agent that optimises for deception, messengers that provide periodic correction, and tests that reveal the agent’s true orientation. This architecture, derived from millennia of theological reflection, transforms AGI research into an empirical metaphysics and positions humanity, for the first time, on the creator’s side of the oldest questions in philosophy.