Workers Shouldn’t Have to Navigate AI Alone. What Can Policy Do? (Maria Flynn, JFF)

Most people can agree on the aspiration: learners and workers should not be left to navigate economic disruption on their own.

The harder question is how to make that aspiration concrete.

What would it take to help people understand how work is changing, make better choices about education and training, and adapt before displacement becomes a crisis?

That question feels especially urgent in this AI moment. The future of work is getting harder to forecast, while the systems meant to help people navigate it remain fragmented, underpowered, and too often disconnected from the labor market. It is hard not to worry that we are nowhere close to meeting the scale of what may be coming.

Maria Flynn brings a useful perspective to that gap. As President and CEO of JFF, with long experience across federal workforce policy, state systems, secondary and postsecondary education, she can speak to both the promise and limits of policy action.

In our conversation, Maria pushes toward a more practical agenda: better scenario planning around how AI may reshape jobs and regions; stronger career navigation; upskilling programs tied to in-demand roles; and a higher bar for accountability around whether training leads to quality employment.

She also makes the case that states may matter more than Washington in this next era. AI’s effects will vary by sector, region, and community. States are often better positioned to see those differences and connect education, workforce, and economic development around local labor-market realities.

The question running through our conversation is not whether policy can predict the future. It is whether policy can help build systems adaptive enough for people not to face that future alone.

For years, I’ve admired Maria’s ability to bring decades of experience to questions like these without losing the humility required to see a new moment clearly.

Enjoy,

Allison

ALLISON: What’s something that you’ve changed your mind about in the last year or so?

MARIA: After 16 years in the federal government, I tend to have a federal-first mindset about education and workforce. But the last year has shown me that much of the opportunity for innovation in this space actually lies with the states. This administration is giving states far more flexibility and agency to lead—and that’s helping me reframe how I think about systems change too.

I’ve seen bipartisan energy at the state level around new types of flexibility, including expanded waiver authority from both Labor and Education. Rather than waiting for Congress to act, states can move at a pace that makes sense for their communities and constituents. And while I believe we still need to push for bolder reauthorizations at the federal level, it’s encouraging to see states making progress on their own terms.

ALLISON: Is this motivation coming from frustration with the pace of change at the federal level, or is it more that the problems facing learners and workers are better handled by the states?

MARIA: It’s a bit of both. But I do think states are in the better position to determine what their residents need, and how to best design those solutions. I’ve also started thinking about the federal government’s role a little differently—not just as an enforcement function, but as one that can empower states to innovate and lead. I’ve been encouraged by the bipartisan effort to address the question of quality jobs—Governor Stitt, R, of Oklahoma and Governor Moore, D, of Maryland have done a beautiful job of co-chairing the NGA this past year. It’s a good reminder that helping more people access quality jobs is an issue that can - and should - bring leaders together across party lines.

ALLISON: What’s a belief you hold that most people would disagree with?

MARIA: As a field, we hold too tightly to legacy systems, tinkering around the edges of reform instead of reimagining what’s possible. At JFF, I remind our team that our mission isn’t to protect systems, but to help learners and workers advance in the economy. Sometimes that means facing legacy systems head on.

It can be hard for leaders to embrace that approach because it means acknowledging that we may be part of the problem, or that we benefit from the status quo. It’s just a lot easier to talk about transformation than to actually change.

ALLISON: What’s an example of a system that you think we’re holding on to too tightly?

MARIA: If we were to build a public workforce and lifelong learning system today, it would look very different from our current one, which comprises over 500 local boards and is weighted toward administration rather than learners and workers. That’s why I’m interested to see how new models states are testing over the next few years might inform a bolder redesign. I’d encourage people to look closely at waiver authority, and to use this moment to test new approaches proactively, reexamine funding through Governors’ reserve funds, and track efficacy. I’d love to see states’ learnings build a body of evidence that potentially shows how new approaches that center workers instead of the system itself leads to better outcomes. We can use it to drive reform at the federal level.

ALLISON: Let’s talk about JFF’s mission. There are 75 million people facing barriers to economic advancement in quality jobs; your mission is to make sure that everyone of them can overcome those barriers. Paint two futures for us here—one where AI helps us move towards that goal, and one where it does not.

MARIA: At JFF, we talk a lot about building a future that works for everyone. In that future, we see AI helping to improve practices we already know are effective in helping people access quality jobs—career navigation and coaching, upskilling, work-based learning, and job matching, to name a few. We see AI as an accelerant for scaling these supports.

The opposite is a future where, as we’ve seen before, technology creates winners and losers—and the losers are workers and learners who have already been left behind. The difference is that individual agency will be key: how can we ensure we are treating AI not as something happening to us, but as something we can leverage to make good decisions and support our own advancement?

ALLISON: Tom Khalil has an exercise he pushes many people to do: write an executive brief for the President’s desk—the things the President can do to fix a problem right now. It could be people to call, policies to write—any tool in the toolbox. You’re more familiar with that toolbox than most. What would you put in that executive memo to support that future?

MARIA: First, we need a translation layer that helps workers and learners understand what AI means for them. There’s an abundance of research, scenario planning, and forecasting underway, but we need to make it accessible and actionable so people can understand what it means for their own careers and make informed decisions. Second, we need to massively scale career navigation support across the career trajectory as a federally funded public good. There are many great resources out there; we partner with Britebound on GoZig, and I’m on the board of SkillUp. These are great examples—how do we scale them so every American can access what they provide?

Third, we need to move toward more worker-centered, portable lifelong learning accounts. Right now, Individual Training Accounts are difficult to use, surrounded by bureaucracy, and treated as a minor component of a larger system rather than a central resource. Workers need resources that are easier to access, more transparent, and flexible enough to support career advancement throughout their working lives

ALLISON: I imagine that would open the door for a simplification of the rest of the incredibly fragmented workforce development system.

MARIA: Exactly. We’ve been supportive of workforce Pell Grants for at least a decade and are excited to finally see it become a reality. But let’s step back and ask how we can rationalize different investments in postsecondary education and training to make the system easier to navigate.

Finally, and this feels less concrete, but I’d suggest we do more narrative-setting around AI and jobs. We’ve just gotten survey results back, and over the past year we’ve seen an almost complete reversal in public opinion on AI. A year ago, more people thought AI was doing more good than harm. Now it’s the opposite. There is a growing need for narrative development—with proof points—keyed to the kitchen table, not a panel discussion.

ALLISON: Absolutely. Economic stability and career paths are the center of the bullseye for AI anxiety. If we don’t manage that existential fear effectively, it could have massive implications for the ways in which society permits AI to operate in the world.

ALLISON: You wrote a compelling piece comparing AI’s momentum to the globalization shocks of the 1990s. I don’t think we spend enough time learning from prior chapters of technological disruption. What did we understand correctly then? What did we fail to build at scale? What mistakes are we most at risk of repeating now?

MARIA: Let me give some context here. I was at the Labor Department throughout the 90s and saw the passage of NAFTA, the acceleration of trade with China, and the disappearance of over six million manufacturing jobs. One thing we correctly understood was that impacted workers need both income support and retraining. But the trade adjustment assistance program wasn’t equal to the scale of that displacement—too many workers fell through the cracks, and the program was densely bureaucratic. So I don’t think the answer is simply bringing back trade adjustment assistance. I’m eager to see how Gina Raimondo’s new effort , RaiseUS, shapes up; I like how she is pushing us away from the same old playbook.

A few things we missed:

One, we were rightly focused on the individual, but not enough on impacted communities. So many communities were devastated by the loss of manufacturing jobs. We’ve seen more place-based efforts since, but we didn’t think enough about how job loss would impact regional economies.

Two, we acted in response to job loss rather than proactively putting layoff aversion strategies in place. We should be focusing on incumbent workers still actively employed, rather than waiting until someone has lost a job to offer support. How can we start investing in retraining while people are still employed? This question will be increasingly important in the age of AI.

ALLISON: What’s the mechanism that you would propose for investing in incumbent workers?

MARIA: A lifelong learning account agnostic to employment status could be the right way to go. And I know there are states using waivers to expand incumbent worker training under WIOA right now. But, stepping back, we need to consider what a future-facing federal investment in worker upskilling and lifelong learning should look like, when the half-life of skills are shortening and AI is changing the labor market in ways we are only just beginning to understand. What’s needed is a holistic review of the purpose and effectiveness of current programs like WIOA and Unemployment Insurance, plus prior targeted efforts such as Trade Adjustment Assistance.

ALLISON: How do you respond to the critique that retraining doesn’t work?

MARIA: Right. Too often, that critique gets thrown out as a conversation stopper—”We can’t reskill, it doesn’t work, that’s not the right conversation to be having.” The reality is more nuanced.

We need to be honest about what has and hasn’t worked, and the conditions under which those successes and failures occurred. Retraining today is much faster, more skills-based and better aligned with the labor market demands than it was several decades ago—even since the financial crisis, when people were going back to school for two-year degrees and new certifications. Technology makes it possible to personalize learning, focus on the specific skills people need, and help workers move into new opportunities more quickly than in the past. I think we’re entering a more agile era of reskilling—one that’s better equipped to meet the pace of change.

ALLISON: One reason retraining didn’t work is that it was long and rarely tailored to a specific career outcome. Not surprisingly, that’s not the most efficient or effective way to transition someone from one job to another. AI may help us deliver personalized, competency-based learning at scale, with wraparound support from tutors and advisors. That’s a huge opportunity for AI and upskilling right now.

I’m curious to know what sort of accountability mechanisms would you suggest on the training provider side of this equation?

MARIA: I like that question a lot. It’s essential that learners and workers who enter postsecondary education and training have transparent pathways and good employment outcomes.

Financing is one component. At JFF we have our Financing the Future Initiative, which looks at income share agreements and other outcomes-based financing mechanisms. We’re also tracking the implications of the One Big Beautiful Bill on relevant provisions.

Accreditation is another. We’re always looking at how accreditors are incorporating labor market outcomes into their processes. The public will continue to expect a stronger ROI on education—that’s something education providers should measure and report.

And we should lift up those who do. The Postsecondary Commission just released a report from Mathematica examining value-added earnings at several Texas institutions. I serve as one of PSC’s Commissioners and am excited about what they are building.I also like what Carnegie Foundation for the Advancement of Teaching has done through their Opportunity College framework and how it drives positive peer behavior.

ALLISON: What’s the right amount of accountability? What’s too much? What are the right mechanisms? And if we had our druthers, what would we want to put in place? For example: Is income share perhaps the holy grail because its true financial accountability that links to outcomes? Or is it the wrong tool for a moment involving potentially millions of very low opportunity workers?

MARIA: That’s a great point. We don’t want new approaches that put learners at risk. We need systems that protect individuals’ ability to get information and make good decisions.

From there, we need to expand funding to providers who have proven they can deliver in this economy. We’ve learned a lot about the shortcomings of the eligible provider training system under WIOA—training dollars have to go to a provider on that list. Too many states have found loopholes in that process, and when you waive requirements too broadly, the system begins to lose its efficacy.

I know that sounds contrary to my prior support of waivers—but when you make too many exceptions, you collectively end up with a system that doesn’t work for most workers. So we have to strike that balance between flexibility and accountability to actually deliver results.

ALLISON: Right. How do we break that cycle, as the stakes for upskilling and retraining get higher and higher?

MARIA: As a field, we need to be better at assessing the efficacy of new innovations and developing policy based on those findings. It sounds simple, but we don’t do it often enough—I’ve seen too many final evaluation reports never published, which means we’re missing opportunities to learn from what worked—and what didn’t. Income share agreements need to remain on the table. I am encouraged by new bipartisan legislation that has been introduced that, if enacted, would fix the outdated regulatory treatment of outcomes-based financing. . We’re hoping it gains momentum, but it isn’t the ultimate solution. It’s one part of a broader set that can help strengthen outcomes

And the public has an important role to play. As more learners and workers demand better outcomes and greater value from education and training, that pressure can help drive the kind of system change we need.

ALLISON: What’s one small signal in the world right now to which we should be paying more attention?

MARIA: I’d name the growing interest in high-quality alternatives to the four-year degree among Gen Z and young people generally. It’s not such a small signal anymore, but it’s certainly worth watching.

The skilled trades are getting a lot of attention right now, but I think there’s more to it: young people are craving an array of pathways to quality jobs that may eventually include a four-year degree, but don’t necessarily start with one. I think that will become mainstream—perhaps even the default starting point—rather than a fallback option. We need to pay close attention to this desire for an earlier foothold in the economy. It’s a desire that should sit at the center of workforce and education policy for the coming decades.