Every Generation Feared the Machine: The AI Jobs Myth

The first sentence every young person hears today is that AI will take their job. I read that sentence backwards. Every generation has lived through the same fear, only swapping the name of the engine. From the inside each wave looks like a mountain with no far side. Yet behind every mountain, new jobs were waiting.

Same fear, new engine

Around 1900, New York streets held roughly 200,000 working horses, and the city choked daily on millions of pounds of manure. Stable hands, blacksmiths, carriage makers, feed sellers and cleaners formed an entire horse-powered order. Then cars arrived and erased most of it within thirty years. Hundreds of thousands of livelihoods vanished, and that is usually where the story stops.

It should not stop there. Cars created assembly lines, mechanics, fuel stations, road crews, motels and roadside diners. Nobody in 1899 could have named motel clerk as the hot job of tomorrow, because the motel did not exist yet. Electricity erased the lamp lighter, the computer erased the filing clerk, the internet erased the travel agent, the smartphone erased the music player. Each time a larger crowd of jobs replaced the ones that left.

Numbers show the far side

A team led by MIT economist David Autor combed through eighty years of census records. MIT News reported the finding in April 2024: more than 60 percent of the jobs Americans do today did not exist in 1940. The Wright brothers wrote down bicycle merchants on the 1900 census form, and airplane designer only became an official category fifty years later. People invent new occupations on top of new tools; they do not discover jobs that were lying in wait.

The bank teller story teaches even more. When cash systems spread through the 1970s and 1980s, tellers were declared finished. Yet economist James Bessen gathered the records: teller numbers in the United States stood near 300,000 in the mid-1970s and approached 600,000 by 2010. The chart published by AEI shows the same picture. Cheaper branches meant more branches, and the teller shifted from counting cash to solving customer problems.

This reversal has a 160-year-old name: the Jevons pattern. In 1865 William Stanley Jevons proved the skeptics wrong when efficient steam engines failed to cut coal use. Cheaper steam was put to more work, so coal consumption climbed. Wikipedia notes the same pattern is now discussed for AI models, and names Satya Nadella and Erik Brynjolfsson among those citing it. When software gets cheaper, more software gets written, not less.

What to do today

In 1930 John Maynard Keynes told his grandchildren that engines would grow so efficient that within a century we would work fifteen-hour weeks with nothing left to do. The full text still sits in the Marxists archive. It never happened. The 1960s brought a presidential commission on automation, the 1980s brought Japanese robots, the 2000s brought offshoring fears. Wikipedia records the same confident line in every wave: this time is different.

My favorite explanation is simple: a job is a bundle of duties. When a system takes a few duties from the bundle, the bundle does not collapse; it regroups. MIT Sloan summarized a Lawrence Schmidt study in October 2025: roles exposed to AI lost no jobs between 2010 and 2023. Where the system covered most duties the role shrank about 14 percent, but where it took only a few duties employment grew. Firms grew faster with the tool and kept people for judgment and fresh ideas.

What does today show? Since generative AI arrived, the jobless rate in the United States and worldwide has barely moved. Quiet hiring sits beside every loud layoff story, and a flat rate means the two roughly balance. Sometimes poor hiring is simply billed to the system. Headlines never cared for 160-year-old economics, and they rarely look past the mountain.

So my advice to young people has three parts, not one. Learn the tool, because whoever masters the tool directs it. Lean into what the tool cannot do: talk through a human worry, untie a hard knot, invent what is missing. And stay ready for jobs with no name yet; in 2019 nobody knew prompt engineer would become a six-figure occupation. Curious people do not get replaced; they get liberated, especially when everyone insists this time is different.