PERSON
Noam Chomsky
The linguist who ended behaviorism by demonstrating that language cannot be learned from data alone—and whose seventy-year insistence on the distinction between explaining a phenomenon and merely reproducing it is the most rigorous standing challenge to the theoretical foundations of modern AI.
Noam Chomsky is the inconvenient witness at every celebration of artificial intelligence’s achievements. Not because he denies the achievements—their engineering value is not in dispute—but because he insists on the distinction the field most wants to blur: between a system that predicts the surface of a phenomenon and a theory that explains why the phenomenon is as it is and not otherwise. In 1959 he reviewed B. F. Skinner’s account of verbal behavior and demolished, in a single long paper, the empiricist program that language could be explained as accumulated statistical regularity.
Large language models—trained on text alone by statistical adjustment across vast corpora—are, in their deepest commitments, the heirs of the position he refuted. That they work as engineering does not, in his view, make them right as science. The whole point of a career spanning six decades and across two quite different intellectual projects—the science of language and the anatomy of