There’s a common fear among people who work in Silicon Valley: snag one of the fast-growing, high-paying jobs in artificial intelligence, or get trapped in the “permanent underclass”, a phrase describing the fate of those who won’t have upward mobility in the age of AI.

The phrase usually describes a future in which AI automates human jobs. But for women in tech, it’s starting to look like the present.

“A lot of people say: ‘Oh, AI is lowering barriers, it’s equalizing the playing field for women,’” said Urvashi Batra, co-founder and CEO of Prioriwise, an AI platform for IT service providers. “I actually think it’s the opposite.”

Batra said people take her less seriously as the founder of an AI company than her male co-founder. When pitching investors, she and her co-founder have learned that they are more likely to get an investment if he does the pitching.

Women made up only about a quarter of new hires in AI roles in the last year, compared with 50% of new hires in non-AI roles, according to a recent report from LinkedIn. In executive roles, that number drops to just 13%. At the same time, LinkedIn’s research shows that women are more likely to work in roles with high exposure to AI disruption, like customer service, meaning they are also at higher risk for job loss due to AI.

AI jobs, whose postings have doubled since 2023, come with a salary premium, on average paying more than twice as much as roles that don’t involve working with AI, according to LinkedIn’s research. But women who do have jobs in AI are disproportionately concentrated in low-paying roles, like data annotators, said Sarah Steinberg, the head of global public policy partnerships at LinkedIn. Across all AI occupations, men have $45,000 higher median pay than women, partly due to the types of jobs they are likelier to have.

“AI is creating some of the fastest-growing, highest-paying and most consequential jobs in the global economy,” said Steinberg. “Women are just strikingly underrepresented.”

The ongoing trend could produce the greatest gender pay gap in generations and leave women out of building a technology that shaped the economy, advocates for women in tech said.

“What we’re witnessing is a sort of backsliding,” said Brenda Darden Wilkerson, president of AnitaB.org, a non-profit devoted to advancing women in tech jobs.

Initiatives to promote women in the tech industry have been ongoing for years, with mixed results. Women now hold about one-third of tech jobs in the US. But research has found that women leave the tech industry at a much higher rate than their male counterparts, citing factors like non-inclusive cultures and barriers to advancement.

‘I’ve seen so many women fall through the cracks’

Women working in AI may face even greater challenges. “AI moves extremely fast, and the pressure to keep up is immense,” said Jayeeta Putatunda, an AI engineering lead at the investment firm Turing. “I have seen throughout my career so many amazing women I know kind of fall through the cracks.”

Putatunda said it was not uncommon to be the only female engineer on a team, and that when women are outnumbered, they have to speak up louder to make their ideas known. Mentorship in the AI field, especially by other women, can be hard to come by. And the breakneck pace of AI advancements means anyone working in the field should expect to put in long hours in order to keep up.

“If you have very ambitious career goals in this field, there will be seasons when you work beyond normal hours,” she said, noting that a culture of working 12-hour days to keep up was not uncommon. “I don’t think there is always a neat shortcut around that than to put in the hours.”

Those kinds of work expectations can make it impossible for everyone to stay in the field. “I went on maternity leave for four months, and by the time I came back, there were completely different frameworks and levels of models,” Putatunda said. Getting caught back up was overwhelming. She said it was only possible with the help of supportive co-workers and a husband who equitably split childcare.

“That is one way women get pushed out of AI,” she said. “They don’t lack the interest or ability to keep up. They may lack the infrastructure that makes keeping up possible.”

The AI industry is so new and fast-moving that it should in theory be more meritocratic. No one has a decade of experience working with a model developed one year ago, which should mean everyone has a level playing field. But instead the field has indexed more on personal networks and other loose signals, according to some women who work in AI.

“Companies are hiring at a breakneck speed, but they’re finding people through the same networks, the same referrals, the same filters that they’ve always used,” Wilkerson said. “Generally, that’s not given women the same sort of exposure they should have.”

When people have tried to address these tech industry problems in the past, the solutions have largely come down to giving women more mentorship and implementing diversity programs with goals around representation. But in recent years, companies have aggressively scaled back their diversity, equity and inclusion (DEI) programs, largely in response to a political climate that discourages them. That has left fewer companies with institutional programs to hire and promote women.

The political climate has made it more challenging to get AI companies to support initiatives focused on women, said Felicia Newhouse, founder of AI Powered Women, an organization that promotes women’s participation and leadership in AI. Newhouse, a former technology product executive with two decades of industry experience, points to the Department of Justice targeting corporations with DEI programs, claiming the initiatives violate anti-discrimination laws. Companies such as Accenture, Deloitte, IBM and PayPal have all agreed to pay multimillion-dollar settlements under this enforcement.

“It is getting harder to go into companies being called AI Powered Women,” Newhouse said. “And our advocates inside those companies are also struggling with having women-focused initiatives.”

Still, she said the risk that women are left behind in the AI workforce was something that kept her up at night. “We’re talking about who captures a major new source of economic mobility,” she said. “The deepest risk is that a participation gap becomes a power gap.”