The Third Magic

The phrase "another industrial revolution" has been used to flatter so many ordinary technologies that it has stopped meaning anything. Smith's piece is interesting because it refuses the comparison, then proposes a much bigger and more uncomfortable one: that the only events in human history big enough to sit alongside the AI moment are the invention of history itself and the invention of science. If he is right, the right question is not whether AI will change the economy but whether it changes what it means to know things.nnThe first magic, in Smith's telling, is the cultural inheritance that lets a five-year-old today wield more practical knowledge than the most brilliant Roman emperor — fire, writing, agriculture, antibiotics, the legible passage of accumulated insight from one generation to the next. The argument is that without this accumulation, every human life would have to start from zero. Most of our daily competence is borrowed from the dead.nnThe second magic is science, which arrives much later and does something history alone cannot: it produces reliable knowledge of phenomena humans had no way to access by direct experience — atoms, viruses, deep time, plate tectonics. Smith's central provocation is that AI may now offer a third magic: the capacity to generate working knowledge and competence even where we lack a clear scientific theory of what's going on. AlphaFold predicting protein structures we don't yet understand mechanistically; large models making predictions whose internal reasoning we cannot inspect. The magic is real, but the epistemology is strange.nnWe picked this because it is a rare AI essay that lifts the debate out of next-quarter productivity gains into civilisational time. For readers who feel the current AI moment is being explained badly by both hype and doom, Smith's frame is genuinely clarifying — and the unsettling part is that he isn't sure how the third magic ends.