Decades before Siri or Alexa or ChatGPT, there was a real person, almost always a woman, at the center of the information world. She was so vital, most long-distance communication could not happen without her.
She was the telephone switch operator, and at one time she was part of the single largest occupation for white, U.S.-born women under age 20.
Today, that job doesn’t exist.
The world is struggling to figure out what the dawn of artificial intelligence means for society, culture and the economy. A lot of ordinary Americans are just wondering what it means for their jobs.
The history of technological progress — from automated telephone switches to the introduction of the washing machine and the personal computer — can offer some guidance. And that guidance suggests women are likely to be among the most affected by AI.
“AI is focused on routine tasks and numbers and finding patterns, and those types of jobs are the clerical, administrative jobs that a lot of women hold,” said Ariane Hegewisch, a senior research fellow at the Institute for Women’s Policy Research. “It’s still 90% of secretaries and clerical workers who are women.”
New research backs this up. The International Labour Organization in March concluded that nearly one in 10 women in high-income countries works in occupations at the highest risk from AI.
Female CDC telephone operator seated at a switchboard, and wearing a traditional headset, reaching for a telephone patch cable, 1974.
Smith Collection | Gado | Archive Photos | Getty Images
The invention that made operators obsolete
In the 1940s, nearly 95% of America’s telephone operators were female. It was a relatively well-paying job for women at the time, and a complicated one: It took months of training to be able to handle the several hundred calls an hour that could come in at peak times. Operators were expected to be speedy, precise — and polite.
These women were hit hard when AT&T introduced the automated telephone switch. It was one of the largest automation investments in history — a close analog to the massive AI investment now underway — and created one of the biggest technological shocks the nation had ever seen.
In the decades after the automatic switch was popularized, an estimated 75,000 operator jobs vanished.
Economists James Feigenbaum from Boston University and Daniel P. Gross from Duke University’s Fuqua School of Business studied census and paycheck data from the time and followed local employment trends along the path of AT&T’s rolling investment in switches across the nation.
A group of female telephone operators for the CDC in front of a traditional switchboard, 1974.
Smith Collection | Gado | Archive Photos | Getty Images
“The effect of this shock on incumbent operators was to dispossess many of their jobs and careers,” they wrote in 2020. Many switchboard operators wound up in lower-paying jobs — or dropped out of the workforce entirely, they found.
Economist David Autor says this outcome reflects the reality that many workers are already in their highest and best-paying occupation. When the job goes away, the next step is often down.
“Most people are not changing careers very often. They make their career decisions early, and they stick with those careers,” he said. “If those careers, if the value of that work … suddenly goes out of fashion, becomes technologically obsolete, they’re going to, in general, do something not as good.”
But Feigenbaum and Gross also discovered that the next generation of women fared better.
“Future workers found work in other fields, often taking up similar-quality jobs,” they wrote, as the younger workers were “more adaptive to an evolving labor market … Anxiety over the opportunities available to future generations proved to be somewhat misplaced.”
In other words: One type of job went away, but other work emerged over time that was just as good.
Portrait of an unidentified model who poses in front of a washing machine and dryer with an armload of solid-colored, folded towels, 1950s.
Tom Kelley Archive | Archive Photos | Getty Images
‘Engines of liberation’
By the 1950s, another technological innovation would come along and have a profound effect on women: the washing machine.
Dubbed “engines of liberation” in a seminal paper on the topic by University of Pennsylvania economist Jeremy Greenwood and others, washing machines and other appliances dramatically reduced the time women spent on unpaid housework. That meant time could be reallocated to a paycheck outside the home, even as women’s responsibility for the home remained.
From 1950 to 1980, as gender norms changed, the number of employed women in the U.S. grew by more than 25 million. Their participation rate, or share of the female population in the labor force, grew to 52% from 34%. At the same time, 16 million more men found work, though the male participation rate dropped by 5 percentage points.
Read more CNBC economics coverage
The relevant question for the AI boom today: How did the economy suddenly find jobs for these women? If AI creates a vast new pool of available workers, can the economy put them to work?
First, the very entry of women into the workforce meant demand for tasks once done at home moved into the paid economy, services like childcare and food preparation. Second, as the nation grew wealthier, other employment opportunities for women emerged in the service economy — in retail, restaurants, leisure and hospitality, and even some professional services. And third, as American businesses grew, they needed more people to handle the paperwork — everything from payroll to customer accounts.
Changes to the economy and businesses were partly related to the entry of women into the workforce, but it was also partly random luck that the U.S. was booming and businesses were changing to need more office workers.
“Between 1940 and 1980, there was a ton of new work created in the U.S. economy, and it was in office and clerical work that was accessible to people with a high school education,” Autor said.
University students typing in computer laboratory.
Claflin University | Historically Black Colleges & Universities | Getty Images
The rise of the PC
The next big revolution affecting women in the workplace was the personal computer.
Before PCs arrived on the scene, women were the backbone of office infrastructure. According to the 1980 U.S. census, nearly 4.7 million Americans worked as secretaries, stenographers and typists. More than 98% were women.
When IBM introduced the personal computer in 1981, followed by the advent of Microsoft Word just two years later, some of the middlemen, or in this case, middlewomen, became obsolete. It didn’t eliminate writing, filing or editing. It simply shifted a big chunk of those tasks from dedicated clerical workers to others in the office — executives could type their own reports, book their own travel and print their own papers.
But the PC wasn’t entirely bad news for women in the workforce. Studies show women were already more likely than men to be using a computer at work at the dawn of the PC era. The PC put new tools at their fingertips: spreadsheets and databases and general computing power. While many roles were eliminated, many women adapted and found new ways to use the emerging technology to make themselves indispensable to their employers.
That’s a potential outcome for AI. In some cases, it will eliminate tasks but not entire jobs, enhancing the time and ability of a worker to do the remaining value-added jobs that only humans can do.
A 2022 paper by Marcus Dillender and Eliza Forsythe found that when companies adopted software to do office and administrative work, it resulted in a 1% decline in employment in the sector but a 3% increase in wages for college graduates.
Many more women have earned college and advanced degrees, and hold managerial jobs, than when other technologies ripped through the economy, offering some potential protection from automation. But maybe not for long: One of the promises of AI is that it will move up the value chain of the jobs performed by humans.
Whatever capabilities the new technology may have now, the outlook is unclear because, Hegewisch says, somewhat ominously, “AI can learn.”