What Deindustrialization Did to Men, AI May Do to Women (Molly Kinder, Brookings)

“Most of the millions of people about to be displaced are women with high school degrees.”

Molly Kinder is a Brookings researcher who studies technology, work, and workers. And yet this year, she has gone to the Vatican not once but twice.

That detail says something about how she sees this moment. We are not lacking analysis and predictions. We are drowning in them. What we are missing is wisdom, the kind that helps us decide what matters, not just what is changing.

Molly has a rare mix of rigor and moral seriousness, and she is becoming one of the clearest ethical anchors in the AI and work conversation. Lately, she has been making two moves that strengthen that role.

First, listen to how she talks about who gets hit first. In our conversation, she names a risk that is both clarifying and unsettling: among the jobs most exposed to AI’s automation risk are millions of clerical and customer service roles held by women with a high school education. GenAI could be to high school educated women what deindustrialization was to high school educated men. She makes it concrete. Think HR assistants, legal secretaries, payroll clerks, medical coders.

These are not throwaway jobs. Many are stable, living wage roles. They can be done from home. They are gentle on the body. They offer a ladder. And yet women in these roles are barely in the story we tell about AI and work. Molly is asking the question that should be central and somehow is not: what happens to women, and to families, if these jobs disappear and the only thing left is low-quality work?

Second, she is pulling her attention away from the release cycle. Each new model, functionally, tells us the same thing: more capability, faster. So rather than chasing the news, she is going back to enduring classics. She is reading Benjamin Franklin’s memoirs to understand work’s role in our nation’s founding, de Tocqueville on American work ethic, papal encyclicals on the dignity of work, Studs Turkel’s “Working”, and the Bhagavad Gita for a non-Western, non-Christian perspective. Across those pages, one theme keeps surfacing: human history treats work as a major source of purpose. Which raises the question underneath all the headlines: what happens if large swaths of the population are without it?

I could talk to Molly for hours. I hope you find her as clarifying, and as productively provocative, as I do.

- Allison

This conversation has been edited for clarity and length.

---

On Wisdom, the Catalyst for Pro-Social AI Innovation

ALLISON: What is something that you’ve changed your mind about in the past year?

MOLLY: I’ve been reconsidering what the world needs in terms of my expertise. In the past two years, scholars like me have been hustling to unpack every change, and we’ve been constantly looking forward, trying to predict the future accordingly. But our focus can’t stay on the little details of every release, because directionally, every release is going the same way: AI is getting more powerful.

We do need more wisdom about what we’re going to do about it. That means drawing more from our past to guide our future – and I believe we need to be seeking that wisdom in a very patient, careful, and interdisciplinary way.

ALLISON: What moment made you feel like you needed to shift your center of gravity from keeping up with the AI Joneses to thinking about new sources of wisdom and guidance for the sector?

MOLLY: I was fortunate to visit the Vatican twice last year for AI conversations, and both times I was inspired by that institution’s wisdom on AI and the future. The essay Antiqua et Nova: What is Old Is New, written by two Vatican-connected philosophers, is just brilliant: it calls us to build technology that sustains the “dignity and vocation of the human person” – and it defines what that means in a multi-faceted, considered way. That prompted me to explore Pope Francis’s writings on technology, then construct a broader historical view of Catholicism’s definition of work, back to Genesis. Its opening pages describe Adam and Eve being put on earth to “till the land.”

In reading these works, I realized that what we’re experiencing will be economically, societally, and personally earth-shattering, on a global scale. We need to ask ourselves: What does work really mean to us? What do we want AI to do for us, for our world?

You don’t need to be Catholic or believe in God to find wisdom in this work or ask these questions. We need many voices and perspectives digging into diverse traditions: I’m reading Benjamin Franklin’s memoirs to understand work’s role in our nation’s founding, de Tocqueville on American work ethic, academic books on a Protestant “theology of work,” and the Bhagavad Gita for a non-Western, non-Christian perspective.

We need to understand what work means to us, and what it has meant to society for centuries, so we can create a future of work upholding our dignity, purpose, and humanity. What should we vow to protect? What needs changing to support human flourishing? We can’t begin asking these questions while chasing the next shiny object. We need stillness and patience to explore, read, think, and have conversations like this one.

Relatedly, we must acknowledge the community building these tools is too narrow; Silicon Valley needs input not just from technology and economics, but from diverse knowledge bases crossing social sciences and humanities. These domains are critical to this conversation, but not currently at the table. We need more wisdom, from more people and disciplines, to navigate this uncharted territory.

ALLISON: One of my strongest convictions heading into this year is that we wholly lack what you’ve called wisdom and what I call positive visions for our future. For good reasons, most narratives are fear-driven. We’re afraid of job dislocation, loss of meaning from work, ensuing inequity, and data centers gobbling up energy without sustainable sources coming online.

Fear gives us guardrails and is essential for being pointedly clear about threats we need to mitigate. But fear doesn’t give us anything to aim for. We need more wisdom informing that vision, informing our design. Otherwise, we’ll simply follow commercial motivations while scrambling to address fear-driven backlash from citizens. That’s not a world I want to live in as our society undergoes immense change.

MOLLY: I agree. But I’m concerned about our ability to create positive visions because, frankly, AI is not a shared project, and Silicon Valley’s point of view is too narrow. Its endgame – replacing work and transitioning to an abundance economy – massively misreads what matters to people: meaningful work, economic dignity and stability, community and connection, the daily discipline of a job, the sense of opportunity for our children, even ambition, striving, and excellence.

For Americans to invest in and embrace a positive future vision, they must trust that Silicon Valley cares about what they care about. That trust doesn’t exist right now. If Silicon Valley wants Americans to buy into dramatic disruption, people must believe everyone will actually share the bounty.

By the way, this isn’t just about getting the AI industry to be better salespeople. I’ve heard chatter about how poor the AI community is at selling their vision, but I don’t buy it as the real problem. We need a wise, participatory conversation about what we’re optimizing for. Even if Silicon Valley, perhaps with help from Washington, releases an absolutely awe-inspiring vision for AI and humanity, our citizenry will fear AI if they are not included in the process. For many Americans, nothing less than agency – freedom – is at stake.

Is this too dark, too dystopian, Allison? It is my honest take.

ALLISON: No, not at all. Systems do what they’re designed to do, and venture-backed systems aren’t designed to manage large-scale social change centered on equity. By expecting them to, we’re actually giving them more power than they deserve. Where do we want frontier model companies’ responsibilities to start and stop? What can they actually do effectively on behalf of the public? They must be told where their lane ends and democratic governance begins.

MOLLY: That’s exactly right. The decisions frontier model companies are making now will impact all of us on a deeply personal level – how we make a living, provide for our families, our children’s well-being and future flourishing. Nothing is more existential than those questions, and AI touches all of it. Right now, our government is not proactively steering AI in a pro-human direction. We must equip our policymakers, our institutions and ourselves with the wisdom and courage needed to confront this change. The stakes are incredibly high.

---

On a Future That Supports the Growth of Every American

ALLISON: To manage toward a better future, I’ve been asking guests to lay out two visions: what happens if we get AI right, and what if we don’t. Could you give me yours? What does each vision – pro-social and dystopian – feel like for the average American worker?

MOLLY: 2026 is my year of solutions. In think tanks, we can spend all day admiring the problem, being the sense-maker for businesses, policymakers, unions. But studying the problem isn’t enough for me. My goal this year is to confidently bring forward solutions to the challenges AI poses to work and workers. .

Last year, along with colleagues from Yale Budget Lab, I published research on the early career ladder concluding that we still can’t determine if AI is driving higher unemployment for young people. But we shouldn’t wait for perfect data to start developing solutions. What if AI is playing a role - or soon will? What should we do? That’s the attitude we need: concrete solutions policymakers can champion and employers can adopt, ready to put into action.

Here’s the future I would love to see: I want AI to increase workers’ capabilities, not substitute for them. The north star is for everyone to be able do meaningful, creative, challenging work. I’d like AI to be designed as a mass product empowering workforces beyond the power users, who are already among the most privileged. Work would still exist, and while jobs would change, we would create new, exciting opportunities. The benefits of AI would extend to the least educated to prevent a deeper divide between those with opportunity and economic power and those without. Importantly, our society would have mechanisms to share productivity gains so benefits don’t accrue only to capital holders. Humans wouldn’t be excluded from economic activity, getting a check and a hobby later – we would be part and parcel of an AI-augmented economy. This is the humanist vision.

If AI goes wrong, our economy would mirror deindustrialization, amplified. During deindustrialization, we lost 8 million manufacturing jobs over three decades – a modest percent of the workforce. We created five times more jobs during that period, but still, the toll on workers and communities was enormous. The jobs people wanted were gone. AI could amplify this scenario many times over. What if AI destroys coveted jobs across the economy? What happens when the best, most upwardly mobile jobs disappear?

This scenario would enable a mass race to the bottom for low-wage jobs. Maybe some jobs would be created, but there would be a mismatch between skills and also interest. Just because a data center requires HVAC techs during the build doesn’t mean displaced bank tellers or market research analysts or copyeditors would want or can step into that work. The goal shouldn’t be to only create net jobs; we will need jobs meeting the preferences, places and talents of people seeking new roles.

That dystopian future happens if we pursue automation intending to replace human labor, but also if we lack a safety net to catch people, help them transition, and secure shared economic benefit from automation.

---

On Women and the Future of Work

ALLISON: Let’s take that context and zoom in on the population for whom you’ve been sounding alarms: women without college degrees. This group is more likely to be impacted by the first wave of automation because they’re disproportionately in roles requiring rules-based, digitally-mediated work. Tell us what’s happening and what comes next for women at work.

MOLLY: The headline for this research, which I’ve been conducting over the past year at Brookings, is that GenAI could be to high school educated women what deindustrialization was to high school educated men.

We all know how poorly deindustrialization went for men. The best jobs for men with high school degrees outside of the trades were in manufacturing. Those were middle-class jobs providing identity and economic stability. They were lost en masse, with most men pushed into very low-quality jobs by comparison, like retail warehousing, or out of the workforce entirely. It was devastating for those communities.

I see potential for something similar here. In an earlier Brookings paper, we argued that GenAI is “is not your grandparents’ automation;” GenAI isn’t impacting physical work, like manufacturing, as it did in deindustrialization. Instead, it’s impacting office-based and computer-based cognitive work. And what we document in our research is that women are more “exposed” to GenAI than men. In particular, millions of jobs held by high school-educated women in clerical and customer service roles are especially vulnerable to automation.

Think back-office roles like HR assistants, legal secretaries, payroll clerks, medical coders. Medical coders and billers make almost $50,000 annually with a high school degree, often from home; some employers I interviewed hire new moms with babies wanting to work at home 15 hours weekly. What other job can you do with a high school degree and a baby at home for the equivalent of $50,000 yearly?

These are good jobs. They’re gentle on the body. They’re not physically demanding. You can retire into them. They have set schedules. Some can work from home, so they can be great for raising a family. They’re more likely to offer a living wage or close to it. And they have upward mobility, unlike low paying jobs in retail, fast food, and home care.

What happens to women in our country if these jobs disappear? Their jobs are the “low-hanging fruit” that everyone describes when discussing the prime targets for AI automation. Without a real philanthropic and policy strategy, I fear that women with high school degrees will have few opportunities outside of low-quality jobs.

It’s not a stretch to say that this would be devastating to the economic security of American families. These women are geographically distributed across the country in communities of all sizes. They’re almost evenly split politically, and many are raising children. But I don’t see a foundation or governor anywhere talking about this; there are no New York Times articles, these women’s stories aren’t in our zeitgeist. Meanwhile, the big investments the country has made over the past decade in job creation have been in physical roles typically filled by high school educated men: trades infrastructure, data center builds, apprenticeships.

This poses an overlooked risk to American women and their families, and we need bipartisan policy solutions now.

ALLISON: What would be a more positive pivot? What categories of jobs would you propose pathways and programs for these women to upskill and move into?

MOLLY: Healthcare offers a big opportunity, which you know well, Allison: it won’t disappear, and we’ll need more as AI helps cure diseases and people live longer. There’s a positive vision here. AI could be key to retraining women into good healthcare jobs. Think how helpful the right AI training could be for a working mom wanting to be an RN, but unable to pursue a brick-and-mortar degree. We just need to ensure that however women upskill and retrain, they get paid for it and AI doesn’t become a race to the bottom. We need to think deeply about that.

There’s also no way around it: we need to make today’s low-wage, AI-resilient jobs better. How are we not completely revolutionizing the value workers can bring through AI, beyond what’s currently possible? This is something I have been working on, figuring out how to harness AI as a “genius and coach” in workers’ pockets to improve jobs. For instance, we are working on a pilot to make home care work better paid with more clinical tasks, enabled by AI, going well beyond (low paid) manual work – and in doing so, put a clinically-trained care worker in every aging adult’s home. We should be doing this broadly across the economy.

---

On The Early Career Ladder

ALLISON: Given that you’re shifting into solutions mode in 2026, what are your thoughts on helping young people build skills for the new “entry-level” – roles requiring confident decision-making, trust, and leadership skills traditionally taught on the job? How do we mitigate what we’ve been calling the experience gap?

MOLLY: I think that’s great phrasing – the experience gap. To reiterate, we have not conclusively proved that AI has destroyed the early career ladder. It is my fear that it will, but we don’t know for sure that this is happening. And, again, we can’t wait for perfect data to start making good plans to prepare if it does.

I like to frame it this way: AI presents both headwinds and tailwinds to young people. I do think it gives superpowers. Workplaces are revolutionizing how they work, fueled more by AI agents, and an AI- savvy young person might be in exactly the right place at the right time to just leap in and be incredibly instrumental.

That said, much entry level work in knowledge sectors that is routine and computer-based is vulnerable to automation. That forces us to reconsider what skill, knowledge, and experience an entry-level candidate really needs.

For me, the new entry level role will carry two expectations: First, young people will need to have the wisdom and judgement to fully utilize AI tools. Second, if the early career tasks that train young people are done by capable AI agents instead, young people at the start of their careers will need entirely new ways of becoming experts.

Ethan Mollick had a great Substack out recently, on teaching his entrepreneurship class at Wharton to work with AI agents. These are mostly executive MBAs who already have subject matter expertise. They’ve already worked out in the real world. They have experience. They’ve managed before. They understand their domains, and thus they know how to ask the right questions or direct these agents to do X, Y, or Z. They knew how to judge the content. How are we going to help new graduates get this level of knowledge and experience?

As AI disrupts white collar industries, we can learn from the sectors where new hires are expected to be experienced experts on day one. Doctors, teachers, dentists, therapists, and nurses are not fully licensed to work until they are practiced, experienced, and able to make high-quality, judgement-driven decisions. They develop this expertise through structured, mentored, on-the-job training. White collar sectors that have historically relied on entry-level grunt work – slide decks, drafting, financial analysis on spreadsheets – may need to look more like these “day one expert” careers in the future. How can we realign career preparation so people can step into demanding jobs employers need to fill on day one?

That’s my diagnosis; now we need a new model and new incentives. I have proposed the idea of adapting the medical residency model to white collar industries. Hospitals have used structured residency programs to enable trainee doctors to develop expertise quickly, under the close supervision of a senior doctor. Knowledge sectors –- from finance to law, consulting and tech - should do something similar.

Here’s why. Learning and hands-on training are expensive. Without intervention, we risk passing the cost of this expense to young people while employers use AI instead to save a buck. If knowledge work industries turn to AI for tasks and hire day-one experts en masse, students will likely be stuck paying for extra training and work experience themselves. Some might work for free, or take out enormous debt for graduate school or professional training.

I would like to see a solution where everyone has skin in the game and the costs of keeping the doors open to the next generation is not borne solely by young people.

What would that look like? Employers would recognize the need for talent in the long run, so they buy into the hands-on training and mentored development that is needed. Schools help meet that need by investing in experiential curriculum and working seamlessly with employers to provide job and internship placements. Importantly, students don’t have to go into deep debt just to participate. We’ll need to harness government and philanthropy to incentivize employers to make short-term earnings cuts for long-term efficiency and profitability. (I pitched the idea of an “AI workforce reinvestment fund” to do just that.) How can we put public money and infrastructure to work creating this training apparatus and compelling companies to participate?

To restate your point, Allison: companies need external motivation to act in society’s interest here because they’re not set up to drive broad social change. We must ensure companies don’t let profits grow at the expense of their future talent pipelines. If we don’t, there will be huge social upheaval – and Americans’ dwindling trust in the corporate sector will be decimated.

ALLISON: Molly, thank you for this inspiring conversation. I love that 2026 is the year of solutions and wisdom. Truly, it gives me such energy to hear you grapple with these problems; thank you for modeling how to situate what’s possible within a broad understanding of the past. I agree – it’s an approach that is desperately needed.

---

Explore more on AI and the future of work:

Andrew Yang on Work, Dignity, and Economic Agency

The Bot Sandwich Economy: Why AI Will Make Some Jobs Better and Others Worse

Can AI Be Pro-Worker? A Former Labor Secretary Makes the Case - Seth Harris

---