I still talk with former students from Carnegie Mellon University and the University of Wisconsin-Madison, including people now a few years into their careers.
I remember the students who stayed after class with one more question, came to office hours with another draft, and seemed determined to squeeze every opportunity out of college. Many were working toward the same bargain:
Learn. Get good at something. Build expertise. Get noticed. Move up.
I admired that ambition. Sometimes I wanted them to temper it—take a class simply because it interested them or spend an afternoon with friends.
But I understood the pressure they felt. Becoming capable was supposed to be the path to opportunity. Now those students are entering a labor market where AI can help early-career employees do work that once required far more experience.
There is a popular story about early-career workers that casts them as disengaged or unwilling to work hard. Those are not the young people I know. The students I taught were ambitious, smart, and intensely adaptable. When the rules changed, they learned the new rules.
In fact, early-career workers may be adapting faster than employers realize. Jobs for the Future’s 2026 research found that workers with fewer than three years of experience were more likely than their more-experienced peers to say they understand AI well. Yet among workers with less than a year of experience, 70% said their employer had never consulted them about how AI should be used in their work.
That is the tension hiding inside AI’s democratizing promise: Young workers are not simply asking how quickly they can learn AI. They are trying to understand what all this learning will allow them to become.
That tension also surfaced in my recent Seramount research, which combines interviews with senior leaders and more than 10,000 qualitative and quantitative responses from Seramount’s Employee Voice Sessions.
Across both sources, I saw that organizations are getting better at helping people adapt to AI. They are less prepared to answer what happens when people become more capable as a result.
The promise is real. That’s what makes the problem harder.
AI can make sophisticated work more accessible earlier in a career.
In a study of more than 5,000 customer-support agents, generative AI increased productivity by 14% overall and by 34% among novice and lower-skilled workers. The researchers found evidence that less-experienced workers adopted AI practices used by more-experienced colleagues.
That is a genuine democratizing promise. But capability is not the same as opportunity.
Organizations increasingly ask whether employees are ready to use AI. What about whether organizations are ready for what employees can now do because of it?
Taylor Sullivan, Vice President of Product at Workera and an expert contributor to our recent research, makes a useful distinction: “AI transformation should not be the outcome. AI transformation should be the means to an end.” Leaders, she argues, need to know what business outcome the capability is supposed to enable.
The same is true of talent. Training employees to use AI is not a talent strategy.
Capability has to go somewhere: into harder problems, greater responsibility, deeper experience, or new opportunity.
Nehal Jani, Chief People Officer at Space-O Technologies, captured employees’ underlying anxiety about this shift: “They’re afraid of becoming irrelevant.”
The fear is not always about losing a job tomorrow. It can be about becoming better at work without knowing whether that new capability leads anywhere.
If you call your workforce AI-ready, can you answer four questions?
1. Do people know what they are building toward?
AI changes the work, but it can also make the future harder to read.
Which capabilities are becoming more valuable? What experience should employees pursue next? Who helps them make sense of the difference?
Seramount’s Employee Voice Sessions show that career progression is already difficult to navigate. Among employees who identify barriers to advancement, 47% cite structural barriers, and 25% cite unclear progression.
People cannot build toward a future they cannot see.
2. Do people have a real way to build capability?
A course can introduce a tool. It cannot create expertise by itself.
Gina Greenwalt, Director of Culture, People Development, and People Analytics at Twitch, and her team are experimenting with a different L&D model.“It’s people at the edge,” she told me, “and finding them, and helping them bring others along.”
Rather than putting L&D at the front of the room, her team helps expertise already emerging inside the organization travel.
Development may happen in a Slack channel, alongside a more experienced colleague, or after a client call. As AI changes the work through which people traditionally build expertise, organizations need to become more intentional about where learning by doing happens next.
AI may make answers easier to get. It does not replace the people who help us understand why the answer matters. Learning and people leaders must find their AI pioneers and ask, who gets to learn from them?
3. Two people can learn the same skill. Do they get the same opportunity?
An employee gains skill with a new tool. Then what? Do they get a harder problem? A stretch assignment? More client exposure? Greater decision authority?
The problem predates AI. A 2026 OECD analysis identified a “broken link” between skills proficiency and skills use: Highly skilled workers can have limited opportunities to apply those capabilities on the job.
AI can widen that gap quietly. It can increase what someone is capable of doing without changing what they are trusted to do.
And capability does not convert into opportunity evenly. One employee gets the visible project, stretch assignment, or sponsor who recognizes what they can now do. Another simply gets more work done.
Those pathways are already uneven. In Seramount’s Employee Voice Sessions, women of color report the highest ambition but the lowest belief that they will advance to a senior role, and 56% report having no sponsor.
AI may democratize capability. It will not democratize opportunity on its own.
4. Are you using new capability to build your future talent pipeline?
When AI changes a task, leaders can easily focus on what disappeared. Or they can ask what becomes possible next: What are we building with the capability AI creates? A stronger internal talent pool? A new path into higher-value work? People who can make increasingly complex decisions?
There is a difference between making today’s workforce more productive and building the workforce your strategy will need tomorrow.
The point of reskilling cannot simply be greater output today. It should help build tomorrow’s workforce with the people you already have.
Development is not adjacent to AI transformation. It is how organizations turn today’s productivity gains into tomorrow’s talent.
Capability needs somewhere to go.
The next AI divide may not be access to the technology. It may be access to what comes after it: the chance to turn new capability into a career.
I keep thinking about the students I taught. They wanted to become people who could think, make judgments, and take on harder problems over time. AI may change how quickly they can produce, but it does not answer the question that shapes a career: Who can I become from here?
That question now belongs to employers too.
If AI makes your people more capable, can your organization give that capability somewhere to go?
That is a more meaningful test of AI readiness: not simply whether people can use the technology, but whether organizations can turn new capability into better work, stronger careers, and the workforce they will need next.
