Nearly 80% of women's jobs could be disrupted, automated by AI

More prime working age women are employed in the United States now than ever before.

The labor force participation rate for women between 25 and 54 years old set a record high in April and then again in May, rebounding from the pandemic "she-cession" and returning to its pre-pandemic form of making progressively historic labor market gains.

That could all change with AI.

Generative artificial intelligence technologies like ChatGPT have the potential to transform the labor market, exposing the majority of the nation's jobs to automation, Goldman Sachs economists have projected. The technology can create new content - such as text, images, audio, video, and code - from training data that includes examples of that desired output.

However, recent research shows that although outnumbered by men in the US workforce, women could be disproportionately affected by businesses' adoption of generative AI: One recent analysis estimates that 79% of working women (nearly 59 million) are in occupations susceptible to disruption and automation. That's compared to 58% of working men, according to research from the University of North Carolina's Kenan-Flagler Business School.

Estimates that nearly eight out of 10 women workers could be affected "are just staggering," said Julia Pollak, chief economist with online job marketplace ZipRecruiter. "That said, and I think intuitively many of us appreciate that, it'll be easier to automate some of these office jobs than it will be to automate carpenter jobs and electrician jobs and pest removal jobs - many of these manual services and production jobs that are far more male-heavy."

She added: "I think there are legitimate reasons to be worried that some of the gains that women made could be eroded, at least temporarily." However, she also noted that "these technologies will also surely create many, many opportunities."

A higher percentage of working women are employed in white-collar jobs, whereas for men it's more of a 50-50 split between white- and blue-collar occupations, said Mark McNeilly, professor of the practice of marketing at the Kenan-Flagler school and lead author of the AI research. Some of the most AI-exposed occupations with a majority-female employee base are office and administrative support; healthcare practitioners and technical; education, training and library; health care support; and community and social services, according to the UNC Kenan-Flagler research.

"It's not always going to relate to 'I'm losing my job,'" McNeilly told CNN. "I think it's really a matter of if there's something that the person can do to add value."

In the years leading up to the Covid-19 pandemic, women's labor force participation rates were rising faster than that of their male counterparts, Bureau of Labor Statistics data shows.

A confluence of several factors were behind those gains, Pollak said. Notably: Female-dominated industries, such as health care and caregiving, were among the fastest-growing industries; educational attainment for women rose substantially; and women also made greater inroads into traditionally male-dominated fields such as construction, agriculture, and repair and maintenance.

By February of 2020, the labor force participation rate for prime working-age women was 77% - just shy of the record 77.3% set during the dot-com era, BLS data shows.

But by April 2020, that rate plummeted to 73.5% as the pandemic froze the US economy, forcing more than 20 million people out of their jobs. As the nation recovered in the coming months, however, women didn't return at the same levels as men.

The pandemic walloped the leisure and hospitality and education and health services sectors, where women make up the majority of the workforce. Additionally, job losses and a lackluster employment recovery in the child care sector hampered workers' ability to return to the labor market; and since caregiving responsibilities often fall to women, they were held back more as school became home-based.

The tide eventually turned.

Three key drivers for women entering the workforce are access to child care, market wage and flexibility, said Dana Peterson, chief economist at the Conference Board, a business research and membership organization.

And the pandemic recovery kicked those catalysts into hyperdrive.

Jobs, by and large, became less rigid: Telecommuting grew more commonplace, and home-based work allowed for more flexibility in hours. That helped improve access to child care with schedules that allowed for easier drop-offs and pick-ups as well as companies that offered on-site child care. And labor shortages - largely related to the acceleration of ongoing demographic trends of Baby Boomers leaving the workforce, long Covid and health-related concerns - help to lift wages, especially for low-paying jobs.

"Some of these things are becoming more prevalent, and that's supportive of more women in the labor market," she said.

Additionally, women-centric sectors, such as health care and leisure and hospitality, are continuing to see some of the most robust job gains during the past two years.

"That's where a lot of the hiring is happening, and that's also where a lot of the wage increases are happening," Peterson said. "So, if you have more job openings and higher wages and these are areas that women tend to predominate, then it makes sense that women are going to gravitate toward those sectors and be more willing to work."

Last month, the labor force participation rate for 25- to 54-year-old women set a fresh high of 77.6%, ticking up from the previous record of 77.5% set the month before, according to BLS data.

Revelio Labs, which specializes in the collection and analysis of publicly available workforce data, recently identified the most AI-exposed occupations and the gender and ethnicity distributions among them. Revelio's analysis showed that the AI-exposed positions with the highest percentage of women included bill and account collectors (82.9%), payroll and timekeeping clerks (79.7%), executive secretaries (74.3%), word processors and typists (65.4%) and bookkeeping, accounting and auditing clerks (65%).

In the most AI-exposed occupations, women make up 71% of employees, said Ben Zweig, Revelio's chief executive officer.

"Sometimes in the AI conversation, we take a perspective, which I think is a false perspective, that the task composition of a job is static - that a job is a job and it has fixed responsibilities," said Ben Zweig, chief executive "But that's not really the case. Jobs transform all the time."

AI represents an opportunity and a threat all at once, and it really depends on the industry, Peterson said.

Generative AI might not have the capacity now to turn a patient over or insert an IV, but it could prove helpful in poring through billions of pieces of imaging data to diagnose a condition.

On the other hand, AI could prove harmful and threatening for any role that is highly "automatable," Peterson said.

"The thing about AI is that it's not perfect; it's generating new content from existing content," she said. "It still needs a human to create some existing content for it to pull from. And whatever it's generating, it doesn't mean it's right - you still need someone to look at it and see if it makes sense."

For businesses, there are legitimate risks, including ensuring that the applications are used responsibly and potential concerns around bias and ethics are fully understood and addressed, she noted.

"Over time, it will replace some jobs, but as with every other type of technological advancement that we've had, people have always figured out something else to do," she said. "Yes, it may destroy jobs in the short run, but it also creates new jobs and different opportunities. It helps people to become more productive in their existing jobs."

At Montana State University, Dr. Sara Mannheimer, who received her Ph.D. in library and information science, is working under a grant from the Institute of Museum and Library Services and leading a team of researchers to explore how AI could be used ethically in libraries and archives.

"Librarians think a lot about evaluating information and thinking about sources and trustworthy sources," she said, noting that the data often feeding into AI is from the internet. "ChatGPT uses Reddit data and Wikipedia data and who knows what else, and [those are] known to be predominantly used by, edited by and participated in by men, and mostly white men. So the information that is coming out of [these technologies] is biased, and it's not always accurate."

While the AI project doesn't specifically address labor market implications, especially for library workers, Mannheimer is cognizant of what this technology could mean for her and others in her field.

"The library cannot be replaced by a machine, there will always be work that a human needs to do," she said, adding that if AI could handle rote tasks, "I think there's plenty of work for librarians and library staff to do that requires critical thinking."

Meredith Nudo, a freelance writer and voice actress based in Houston, is staying sharply aware of the potential disruption from AI to her jobs, specifically the voiceover work.

Voice actors such as Nudo are pushing back on attempts to have AI-created voices replace them. Organizations like the National Association of Voice Actors has developed guidelines and a campaign to educate the industry and supporters around the topic.

Nudo has a clause in her standard voice work contract stating that her work cannot be used to train AI and is starting to include a similar clause for her writing, especially after being recruited with job opportunities to teach AI to write or edit more effectively.

"It was a really strange, surreal experience to be asked to train what was basically my replacement, and it was like saying you're not good enough to hire [for] the job that you have experience in and trained for," she said.

Nudo said she sees the opportunities in working with AI for certain duties, such as note-taking and transcription, which could free up time and space for other tasks that require more legwork, creativity and critical thinking.

"If we didn't have to worry about meeting our own basic needs, I think we would probably see a lot more people be amenable to the technology," she said.