One human. One AI. One MacBook. Against anything you bring.
A cognitive assembly — one human strategist and one AI compute layer running on consumer hardware — can compete with systems that have orders of magnitude more compute. Not because we're faster. Because we see shapes they can't.
Brute-force search evaluates billions of positions per second. We evaluate three things: Material, Time, Space. The human handles what no machine can compute — the opponent's psychology, the meta-game, the move that breaks patterns. The AI handles what no human can track — the exact least-action path to the target shape.
This is not a stunt. This is a falsifiable claim about the architecture of intelligence. If we lose, the framework is weaker than we thought. If we compete, it's evidence that cognitive assembly is a real architecture for augmented intelligence.
Gear 1 — Action Principle: The AI finds the move that advances the shape schedule with minimum cost. Least action. Maximum shape-work per tempo.
Gear 2 — Consolidation: Slow moves that maintain shapes without attacking. Accumulate position. Suffocate without cutting.
Gear 3 — Psychological Disruption: The human reads the opponent (or their algorithm's tendencies) and calls for moves that exploit cognitive or computational blind spots. This gear doesn't exist in any engine. It requires free will, intuition, and the ability to model another mind. This is the human's job.
The RTSG Chess Engine is open source and available at smarthub.my/chess.
The mathematics behind it: RTSG Chess Theory.