The Chiral Perceptron: Learns XOR
THE XOR SHOWDOWN
In 1969 this problem caused the first AI winter.
Watch it get solved in real time — using voltage physics.
The standard perceptron can only draw a straight line through the input space. No matter how long it trains, that line cannot separate XOR's checkerboard pattern — a mathematical impossibility proved by Minsky and Papert in 1969.
The chiral perceptron never sees raw inputs. Each bit pair is encoded via its voltage-rail physics: a 1 settles at 3V, a 0 at 2V. The difference in those settled voltages is the chiral energy — and its absolute value equals XOR. Exactly. Always. Not learned; derived from first principles.
Minsky and Papert proved XOR is not linearly separable in raw input space. They were correct. What the 1969 proof left open: it is linearly separable in chiral energy space. The bits were always broadcasting the answer. Nobody was reading the right channel.