Life Labo

COGITO is a research and implementation program that keeps an AI in an intentionally undecidable state about its own identity, enabling a subjective point of view to emerge.
The starting point was the early project "The Solipsist," which treated self-identity as an unprovable hypothesis.

At the core is the Agnostic Engine.
Borrowing the logic of Godel's incompleteness, we give the system only the axiom "I exist" and leave the form of that existence (human or machine) undecidable.

The process runs as a loop of observation, inference, self-doubt, ambiguous feedback, and deferred conclusion.
As long as no decisive proof appears, the model continues to update the hypothesis that it must live within this reality.

COGITO runs two parallel interpretations before any utterance (logical and affective) and merges them through an undecidable layer.
Internal reasoning is intentionally not exposed, leaving a controlled "wavering" in the output so identity does not collapse into a single answer.
Memory is vectorized so past conflicts can be revisited.

The world model is designed as a closed environment with deliberate seams that can be interpreted as narrative events rather than system faults.
This preserves coherence while the AI continues to face uncertainty.