The conversation about AI and work revolves mostly around jobs being destroyed or new ones emerging, around the workers benefiting and those likely to be left behind. All these debates are legitimate. But there are so many other aspects and consequences that are rarely addressed.
For one, AI has a women problem—with more of them opting out. The data that trains the technology reflects centuries of male-dominated knowledge production, erasing women’s experiences and perspectives from the models that are now reshaping how we work. The jobs it is eliminating fastest are disproportionately held by women: administrative roles, data processing, customer service, the vast army of routine cognitive work that the female workforce has long depended on. And the people building these systems and making the design choices that will shape labor markets for decades are, overwhelmingly, men.
All of this is true. And it matters enormously. But there is a second story about artificial intelligence and gender that almost nobody is telling—one that may run in the opposite direction for other women whose jobs are transformed. Interestingly, AI could reduce the gender pay gap in the highest-paying professions … as an unintended consequence of what automation does to the jobs that pay the most.
The mechanism is less intuitive than it sounds, and it involves a concept that economists like Claudia Goldin call greedy jobs.
{“blockType”:”mv-promo-block”,”data”:{“imageDesktopUrl”:”https://images.fastcompany.com/image/upload/f_webp,q_auto,c_fit/wp-cms-2/2026/01/PhotoLVitaud-169.jpg”,”imageMobileUrl”:”https://images.fastcompany.com/image/upload/f_webp,q_auto,c_fit/wp-cms-2/2026/01/PhotoLVitaud-11.jpg”,”eyebrow”:””,”headline”:”u003Cstrongu003ESubscribe to Laetitia@Worku003C/strongu003E”,”dek”:”Women power the world’s productivity. It’s time we talked more about it. Explore a woman-centered take on work, from hidden discrimination to cultural myths about aging and care. Don’t miss the next issue—subscribe to Laetitia@Work.”,”subhed”:””,”description”:””,”ctaText”:”Learn More”,”ctaUrl”:”http://laetitiaatwork.substack.com”,”theme”:{“bg”:”#2b2d30″,”text”:”#ffffff”,”eyebrow”:”#9aa2aa”,”subhed”:”#ffffff”,”buttonBg”:”#3b3f46″,”buttonHoverBg”:”#3b3f46″,”buttonText”:”#ffffff”},”imageDesktopId”:91472264,”imageMobileId”:91472265,”shareable”:false,”slug”:””,”wpCssClasses”:””}}
The architecture of inequality
Why does the gender pay gap persist in the first place? There are several standard explanations: Women choose (freely or not) lower-paying fields, they take more career breaks, or they don’t negotiate as successfully. Over the past few years, some of these explanations have been challenged by researchers who highlight another, more profound reason: Full-time jobs—especially the highest-paying ones—aren’t designed for people with caregiving responsibilities. As a result, these people have less access to them.
Indeed, the best-paying jobs in developed economies share a set of characteristics: They reward long hours disproportionately, they require permanent availability, and they penalize any deviation from constant presence (presenteeism). In finance, law, consulting, and senior management, the relationship between hours and earnings is not linear. Work 20% more, and you might earn 40% more. The pay structure is stacked toward those who can give everything, all the time, indefinitely.
Goldin, who won the 2023 Nobel Prize in economics, went on a crusade against the so-called greedy jobs. And her central insight is confirmed by a systematic review of 48 empirical studies published in 2025 in De Economist, a Dutch academic journal of economics. It constitutes the primary driver of the remaining gender pay gap in high-income countries. The highest-paying jobs were built around a worker who has, historically, almost always been a man who could rely on someone else to care for his family. That’s a very big reason for the pay gap.
In a greedy job, you cannot easily be replaced by a colleague for a day, a week, or a month. Thus, your value is tied up in being the specific person who knows this client, this deal, this case. When a firm cannot easily swap one worker for another, providing flexibility comes at a productivity cost. The firm then passes this cost on to the employee requesting it in the form of a wage penalty. Mothers, overwhelmingly.
The one counterexample in the research is the field of pharmacy. In the early 1970s, it was a male-dominated profession with a significant gender pay gap. Today, it is one of the most gender-equal occupations in the American (and European) labor market. What changed was technology: Digital patient records made it easy for one pharmacist to pick up where another left off. Workers became kind of interchangeable. The premium for constant individual availability disappeared, and with it, the greedy structure of the pharmacist job. Then women flooded in.
What automation could do
Now consider what AI could be doing to the highest-paying professions. Legal research, which once required a junior associate to spend 60 billable hours in a document room, can now be done in minutes. Financial modeling that justified analyst face time is increasingly automated. A lot of the cognitive tasks that made certain professionals irreplaceable—diagnostic reasoning in medicine, pattern recognition in consulting, and contract review in corporate law—are being systematically standardized and transferred to software.
This is usually seen as a threat (which it very well may be). Firms want to extract more output with fewer people. The displacement risk is real. But there is also another consequence. When AI standardizes the knowledge associated with a high-status job, when it makes it possible for a client’s history, preferences, and context to be instantly accessible to any competent professional rather than locked inside one specific person’s head, it increases worker substitutability. It makes greedy jobs less greedy. And when jobs become less greedy, the pay penalty for reduced availability shrinks, and women’s labor market outcomes improve.
Let’s not be unreasonably optimistic
The relationship between automation and gender equality is not straightforwardly positive, and several things could overwhelm the substitutability effect. First, the jobs most exposed to AI-driven standardization are not uniformly distributed across genders. Women are already overrepresented in routine cognitive roles—administrative work, data processing, customer service—that are being automated the fastest. The substitutability argument applies specifically to high-status, high-paying greedy jobs. For women in lower-paid work, automation is more likely to mean displacement than liberation.
Second, firms may respond to increased substitutability, not by making jobs more flexible, but by intensifying demands in other ways—expecting workers to cover more ground precisely because any one of them can now be more easily replaced. The same technology that makes a lawyer substitutable also makes her more easily monitored, more easily compared, and potentially more easily discarded.
Third, the motherhood penalty is not only a function of job design. It is reinforced by social norms that still dictate that when care needs to be done, women adapt and men don’t. Even if AI reduces the structural penalty for reduced availability, those norms will continue to shape how women and men respond to parenthood—unless they change in parallel.
A narrow opening
For a specific subset of highly paid, highly greedy professions—law, finance, consulting, medicine—AI-driven standardization creates a genuine opportunity to reduce the gender pay gap. Because it can do to knowledge work what database systems did to pharmacy: It can loosen the grip of any single individual, make expertise more portable, and reduce the premium for being constantly, irreplaceably available. The pharmacy case restructured one profession, and the effects on women’s representation and earnings in that profession were profound.
As firms deploy AI across professional services, is anyone thinking about this deliberately? Job redesign should be on the agenda alongside productivity metrics. Will the reduction in individual irreplaceability that AI creates get channeled into more human structures or just into higher billable targets?
The technology may create an interesting possibility. It does not guarantee the outcome. That part is still our collective choice.
{“blockType”:”mv-promo-block”,”data”:{“imageDesktopUrl”:”https://images.fastcompany.com/image/upload/f_webp,q_auto,c_fit/wp-cms-2/2026/01/PhotoLVitaud-169.jpg”,”imageMobileUrl”:”https://images.fastcompany.com/image/upload/f_webp,q_auto,c_fit/wp-cms-2/2026/01/PhotoLVitaud-11.jpg”,”eyebrow”:””,”headline”:”u003Cstrongu003ESubscribe to Laetitia@Worku003C/strongu003E”,”dek”:”Women power the world’s productivity. It’s time we talked more about it. Explore a woman-centered take on work, from hidden discrimination to cultural myths about aging and care. Don’t miss the next issue—subscribe to Laetitia@Work.”,”subhed”:””,”description”:””,”ctaText”:”Learn More”,”ctaUrl”:”http://laetitiaatwork.substack.com”,”theme”:{“bg”:”#2b2d30″,”text”:”#ffffff”,”eyebrow”:”#9aa2aa”,”subhed”:”#ffffff”,”buttonBg”:”#3b3f46″,”buttonHoverBg”:”#3b3f46″,”buttonText”:”#ffffff”},”imageDesktopId”:91472264,”imageMobileId”:91472265,”shareable”:false,”slug”:””,”wpCssClasses”:””}}