CONCEPT

Predicting Without Understanding

The condition—established as legitimate by quantum mechanics and reproduced in a more troubling form by machine learning—in which a system predicts with extraordinary accuracy while leaving its meaning entirely unresolved: a gap between predictive power and interpretive understanding that Max Born’s physics was the first enterprise to inhabit and that AI has now reproduced at civilizational scale.

Quantum mechanics was, from its birth, the most spectacularly successful predictive theory ever devised and simultaneously the least understood. It computed the outcomes of experiments to a dozen decimal places while leaving its practitioners in deep and lasting disagreement about what it meant—what the wavefunction was, why the probabilities, what was really happening when a measurement occurred. Max Born gave physics the rule that made the predictions work, and even he could not say, to his own satisfaction, what the rule revealed about reality. Generations of brilliant physicists used quantum mechanics with total confidence in its predictions and total disagreement about its interpretation. The theory works whether or not anyone understands why. This is a profound and underappreciated fact: that the most successful theory in science is one its makers do not agree how to interpret. Machine learning

Keep reading with YOU ON AI