Some employers aren’t setting aside time during the workday for any of that, which means whoever has the most time left over after work is quietly getting a head start.
getty
Imagine three people entering the same baking contest.
One has no kids and no one depending on her. Most evenings, she has a free kitchen and a few hours to experiment with a recipe until she gets it right.
Another is caring for an aging parent. Her evenings go to doctor’s appointments, medication schedules and making sure someone else is okay before she can think about her own night. The kitchen is available. The time isn’t.
The third picks up her kids after work, makes dinner, handles homework, cleans up and gets ready to do it all again tomorrow. Somewhere in there, she’s supposed to find time to practice too.
On competition day, the judges see three finished cakes. They don’t ask who had a free kitchen most nights and who was stretched between three other jobs first. They just taste the cakes and rank them. Whoever practiced the most looks the most skilled.
The contest calls that a skills gap. It isn’t. It’s an opportunity gap, and it’s not the same size for all three of these women.
Now replace baking with learning how to use AI at work. Companies want people who can incorporate AI into workflows, experiment with emerging agentic tools, judge the quality of what those systems produce and know where AI can actually make their work better. Getting good at that takes the same thing baking does: practice, repetition and room to get it wrong a few times.
Some employers aren’t setting aside time during the workday for any of that, which means whoever has the most time left over after work is quietly getting a head start.
But “whoever has time left over” isn’t one group. A woman without caregiving responsibilities, someone caring for a parent and a working mother are not competing on the same footing, and they’re probably not falling behind by the same amount either.
Most conversations about this treat “women” or “caregivers” as if they’re interchangeable, one blurry category of people who are somehow behind.
getty
Most conversations about this treat “women” or “caregivers” as if they’re interchangeable, one blurry category of people who are somehow behind. They’re not the same group, and lumping them together doesn’t answer which of them actually has the widest gap, and if it is as wide as we assume.
That’s what makes the University of Phoenix collaboration with OpenAI worth watching. Last month, University of Phoenix and OpenAI announced a collaboration to explore AI across teaching and learning, student support, career services, operations and collaborative research. In interviews, Jamie Smith, the university’s chief information officer, and John Woods, its provost and chief academic officer, said faculty plan to study differences in AI adoption across gender, caregiving status, career stage and industry.
The research could help employers determine whether an AI skills gap reflects ability or simply who had enough time to practice. More importantly, it could answer which group actually shows the widest gap: women generally, caregivers generally or working mothers specifically.
The Skills Gap Isn’t Evenly Distributed
Woods said the collaboration builds on faculty training and AI competencies already incorporated into coursework. Students have logged more than 9.8 million assessed skills and earned more than 1.1 million skills badges, seven of them focused on AI.
Smith described AI fluency less as knowing a particular tool and more as understanding the work around it. Which parts of a job can be sped up? Which parts create privacy or security problems if they’re handed to a model? Which parts still require a person because the decision involves empathy, judgment or accountability? That can only be learned by working with the technology, watching it get things wrong and figuring out what still needs you.
The 2026 Stanford AI Index found AI skills appearing in 2.5% of U.S. job postings, up 55% year over year, while mentions of agentic AI skills rose more than 280%. There’s already evidence of a gender gap in AI adoption for professional purposes. What’s missing is the breakdown that would actually answer the question this piece keeps circling: whether that gap belongs to women broadly, caregivers broadly or specifically to working mothers.
Right now, the data that exists mostly measures gender. It wasn’t built to tell the three bakers apart. The collaboration is particularly interesting in that context because students at the University of Phoenix are largely working adults. According to the university, its average new student is 38 years old, 62% are employed and 70% are women.
For all his optimism about the access University of Phoenix could provide, Woods was refreshingly candid about how this effort could fail.
getty
Where This Could Go Wrong
The first place is the one already sitting inside the research question. If University of Phoenix studies “gender” and “caregiving status” as two separate variables but never crosses them, meaning it never specifically isolates working mothers from women without caregiving duties or from caregivers who aren’t mothers, the findings will describe a gap without ever telling us whose gap it actually is.
Access to a course still isn’t the same as having time to experiment at work. University of Phoenix can give any of these three women the same coursework, the same badges and the same opportunity to practice. And because many students are already employed, researchers can study how those skills show up on the job. What the university can’t control is whether an employer gives a working mother, or a woman caring for a parent, the hour at her actual desk that the hypothesis says she’s missing.
Opportunity might not be the whole story, either. Alongside the access gap, some research points to a trust gap: women who do have access to AI tools still adopt them more slowly, in part because they’re less confident in the output or worried about being penalized for leaning on it. In other words, even a woman handed a fully stocked, fully available kitchen might still hesitate to enter the contest.
And if hiring managers and promotion committees are working off the same signals they always have, the skill gap can close while the career mobility gap stays exactly where it was. The most likely way this experiment fails is in a workplace that never changed what it was paying attention to and never learned to tell these three women apart either.
