Every white-collar worker asking whether artificial intelligence will make their job disappear in five years has been looking at the problem wrong, according to a new essay in The Atlantic. The real question is simpler and stranger: are you coal, or are you a horse?
Staff writer Annie Lowrey has laid out a framework that cuts through the noise of the AI jobs debate, "Hvylya" reports, citing The Atlantic.
The horse side of the analogy is bleak. American farms employed over 26 million equines in 1915. A century later, that number had collapsed to roughly 700,000. Horses did not retrain, did not pivot to factory work, did not learn to code. "They stood there and ate carrots," Lowrey writes.
Coal tells a different story. When steam engines got more efficient, economists expected demand for coal to drop. Instead, cheaper energy opened new uses, drove down consumer prices, and pushed coal deeper into every corner of the economy. Demand went up, not down. The British economist William Stanley Jevons spotted this in 1865, and the insight now carries his name.
The test for any worker, then, comes down to whether technology acts as a complement or a substitute. If AI makes your work cheaper and opens new demand for what you do - the way better imaging created more work for radiologists - you are coal. If AI simply replaces what you do entirely, with no new demand to compensate, you are a horse.
Lowrey acknowledges the framework has limits. Government regulation, corporate decisions, and sheer friction can shift the outcome. The most common job in the San Francisco Bay Area is not AI-systems architect - it is home health aide.