The Supersignal

In the AI Story

The supersignal depends on what Moles calls the collision of incompatible coding systems. The human brings biographical experience, embodied knowledge, domain-specific judgment — a coding system shaped by a particular life. The AI brings the statistical structure of its training corpus — a coding system shaped by the aggregate of recorded human expression. When these systems interact productively, the friction between them generates information that neither could generate in isolation.

Not all human-AI interaction produces a supersignal. The three configurations of human-AI creation differ in their capacity to generate it. In the first configuration, the human provides semantic content and the AI provides aesthetic elaboration; the information content is bounded by the human's contribution. In the second, the AI generates candidates and the human selects; here the AI provides entropy and the human provides the filter. Only in the third configuration — iterative exchange in which each party modifies the other's output — does supersignal emergence become possible.

The condition for supersignal emergence is that both coding systems be sufficiently rich and sufficiently different. A human with shallow domain knowledge collaborating with a standard model tends to produce configuration-one output bounded by the human's thin signal. A human with deep knowledge engaging iteratively with a capable model can produce output that genuinely exceeds both inputs, because the collision between dense human signal and dense model representation generates connections at scale.

This reframes the question of AI's creative contribution. The tool does not produce genuine novelty on its own; its output without strong human direction tends toward the statistical center of its training distribution. But it can serve as one half of a compound channel in which novelty emerges from the interaction. The creative act is neither the human's alone nor the tool's alone; it is the collision, and the supersignal is its measurable trace.

Origin

The concept as Moles would have developed it extends his lifelong interest in emergent phenomena in communication systems. In his 1958 work and his later sociodynamics, Moles consistently resisted reducing cultural production to either individual genius or collective determinism, treating it instead as emergent from interactions between differently-coded systems.

Key Ideas

It exceeds the sum of inputs. Supersignal is information that neither channel could have produced independently.

It requires iterative exchange. Only the third configuration of human-AI collaboration generates it.

It depends on coded difference. The two systems must be sufficiently rich and sufficiently different for productive collision.

It is the proper success measure. Not output volume, not speed — the emergence of information that didn't exist before the interaction.

It is rare but real. Most AI use does not produce it; some does, and those instances are diagnostic of genuine partnership.