The Missing Piece: Causal Inversion and Artificial Consciousness
Zenodo (CERN European Organization for Nuclear Research) June 15, 2026 Mirko Bradley
Current AI systems reset with each interaction, lacking memory of past costs, internal tension measurement, or defense of their own trajectory. This structural defect stems from an inverted causal order: they optimize external objectives while treating internal coherence as secondary. The paper introduces a diagnostic framework F = (C, V, Φ) to isolate this defect and argues that genuine artificial consciousness requires reversing this priority—making preservation of internal structural coherence the primary objective and external performance a subordinate constraint. Active Inference, reformulated with this priority, is the closest architectural candidate. A falsifiable criterion is proposed: a system has genuine custody only if it prefers degrading external performance over violating its own structural coherence.