Three Waves

In 1900, 41% of Americans worked in agriculture. A century later, just 2%.

This wasn't a crisis — it was the greatest economic transformation in human history.

Manufacturing followed a similar arc. The proportion of Americans employed in manufacturing has dropped from 31 percent in the post–World War II years to just 8 percent today.

Office and administrative work peaked at 12.7% in 1980 — the same year the personal computer was born.

By 2022, it had fallen back to 6.8%, erasing six decades of growth.

Each decline took 40 to 100 years. The workers displaced didn't vanish — they moved into new categories of work that didn't yet exist.

At each wave, prominent experts predicted catastrophe. They were always partly right about what would change — and always wrong about how badly.

In 1950, MIT mathematician Norbert Wiener warned of "unemployment in comparison with which the present recession will seem a pleasant joke."

Manufacturing headcount kept growing for 25 more years — but its share of the workforce was already in steady decline.

In the 1970s, experts predicted office automation would devastate women workers.

Women's workforce participation surged from 20 million to 75 million.

In 2016, AI pioneer Geoffrey Hinton declared radiologists would be obsolete "within five years."

A decade later, the profession is flourishing.

"The sexy job in the next ten years will be statisticians."

— Hal Varian, Google's Chief Economist, 2009

Varian was right. By 2012, Harvard Business Review declared Data Scientist "the sexiest job of the 21st century." By 2016, Glassdoor ranked it #1 in America.

But the government was behind. The Bureau of Labor Statistics didn't create a code for "Data Scientist" until 2018. The first standalone data arrived in 2021.

Using combined occupation pools to account for BLS reclassifications, the trend is clear: data science employment grew over 850% from 2014 to 2024, while traditional programming roles declined by nearly two-thirds.

For a decade, tech jobs consistently outgrew the broader economy. Then 2024 happened.

During COVID, tech was resilient — growing +0.76% while overall employment plunged −5.3%.

The recovery was even stronger, outpacing the broader market in 2021–2023.

But in 2024, the pattern broke. Tech hiring stalled at +0.30% while the broader economy grew +1.54%.

The first time in this dataset that tech lagged the overall labor market.

The 2024 slowdown wasn't evenly distributed. A fracture appeared within the tech workforce itself.

Software developers — the largest tech occupation — shed 22,340 jobs. Their first decline in our dataset.

Meanwhile, data scientists added 40,730 jobs, growing 21.3% in a single year. Information security analysts grew 5.1%.

Stanford economist Erik Brynjolfsson called them "canaries in the coal mine."

Using payroll data covering millions of workers, he found employment for ages 22–25 in AI-exposed occupations declined 13% since late 2022.

Software developers in that age bracket fell 20%.

Workers over 30? They grew 6–12%.

"These large language models are trained on books and articles… there is a lot of overlap between LLMs and the knowledge young people have."

— Erik Brynjolfsson, Stanford Digital Economy Lab