The river's first channel is the stable configuration of simple matter. A hydrogen atom — proton and electron bound by electromagnetic force into a persistent pattern — is the simplest possible act of anti-entropic organization. It is not alive. It is not intelligent in any useful sense. But it is ordered, and the ordering is the seed of everything that follows. Stuart Kauffman's work on autocatalytic sets describes the next channel: chemical systems complex enough to sustain themselves through feedback loops, generating what he calls 'order for free' at the edge of chaos. Biological evolution amplifies this by orders of magnitude, producing cells — membrane-bound systems maintaining internal chemistry against environments that would destroy them — and ultimately brains, the most anti-entropic structures known.
Each channel widens the river by enabling new forms of pattern transmission. Genetic evolution transmits patterns across generations at evolutionary timescales. Neural networks transmit patterns across minds at lifetimes. Language transmits patterns across space at conversational speed. Writing transmits them across time at civilizational scale. Each technology is, in Wiener's terms, a new mechanism for maintaining order against entropic pressure — a new channel in a river whose direction has been constant since the beginning: more information, more pattern, more organized complexity per unit volume of spacetime.
Artificial intelligence is the latest channel, and it is opening at unprecedented speed. What makes it different is not that it represents a departure from the biological trajectory — it does not — but that it is the first channel that operates on silicon rather than carbon, electricity rather than biochemistry, and that can process and recombine the accumulated informational wealth of all previous channels at rates that biological cognition cannot approach. The river is not just gaining another tributary; the river is acquiring a new medium through which its patterns can flow.
The thermodynamic question Wiener's framework forces is whether the new channel produces genuine novelty or sophisticated recombination. A large language model is trained on a substantial fraction of recorded human thought. Its outputs are recombinations of patterns already present in the training data — brilliant recombinations, often, but recombinations nonetheless. Is the model adding to the river, or redistributing water already in the channel? The question does not admit a clean answer, and Wiener's framework does not pretend to provide one. What it provides is the vocabulary to ask the question precisely, and the suggestion that the answer may depend on what the human contributes: the purpose, the judgment, the care that transforms sophisticated recombination into something that adds to the order of the world.
Segal's metaphor was developed in You On AI (2026) as the organizing frame for his argument that AI is not discontinuous with the long arc of human tool-making but continuous with it — a new channel in a river that has been flowing for billions of years.
Wiener's thermodynamic framework predates Segal's metaphor by seven decades. The explicit synthesis — intelligence as anti-entropic channel, AI as the latest such channel — belongs to the You On AI Encyclopedia and to the cybernetic tradition Wiener founded.
13.8 billion years of flow. The trajectory of pattern-accumulation is as old as the universe.
Channels, not origins. Intelligence does not begin at humanity; humanity is a recent channel through which intelligence flows.
Each channel widens the river. New mechanisms for pattern transmission enable new kinds of order.
AI as latest channel. Silicon-based computation is continuous with carbon-based cognition in its thermodynamic function.
Novelty vs. recombination. Whether the new channel adds to the river or redistributes is the open question.