Artificial Intelligence, Human Cognition, and Conscious Supremacy

Ken Mogi proposes “conscious supremacy”: tasks that conscious processing might accomplish within practical resource limits that non-conscious processing cannot match. The May 13, 2024 hypothesis-and-theory article uses computational comparisons and literature discussion. The proposed difference concerns practical efficiency, not a computation impossible in principle for a non-conscious system.^[1]

Contribution and alignment relevance

The proposal shifts attention from subjective experience alone to possible functional advantages. Mogi suggests dividing tasks between humans and AI, while acknowledging that restricting AI to non-conscious operations does not guarantee alignment. The proposed division of labor remains conditional on finding such tasks.^[1] This connects Machine Intelligence and Machine Introspection with questions about appropriate human oversight.

Limits

The paper does not identify a demonstrated computation uniquely requiring consciousness. Its suggested conscious error correction and quantum analogy are speculative. Human advantages over existing systems would not by themselves prove that consciousness causes those advantages. The article supports an attributed research hypothesis, not a conclusion that AI is conscious, cannot become conscious, or is safely aligned.^[1]