This week, AI reskilling got a new verb. At Fortune’s AIQ Summit, chief people officers from some of the largest companies in the country said the employees who thrive will be the ones who can unlearn. A few days earlier, McKinsey estimated that millions of American workers may need to leave their occupations entirely by 2035.

I agree with the diagnosis. Unlearning is the right word for this moment.

But I notice which direction it keeps pointing. Down. Toward employees. Toward individual responsibility. Almost never toward the leaders and systems doing the asking.

If everyone has to unlearn, leaders go first.

Key takeaways

  • McKinsey Global Institute estimates about 11 million U.S. workers may need to change occupations by 2035, and only one in seven has a direct path to a growing job.
  • Employers increasingly want people who can learn fast and unlearn, but AI reskilling cannot be handed to employees as a solo responsibility.
  • Unlearning is a leadership task first: the assumptions, career ladders and signals leaders rely on are part of what has to change.
  • The organizations that keep people through this shift will be the ones that define the learning exchange clearly, not the ones that simply supply tools.

What’s happening with AI reskilling

Three stories from the past week fit together.

First, the scale. On September 29, the McKinsey Global Institute published Workforce in Motion: Skills and Pathways to Future Jobs in the United States. It estimates that automation could reduce demand for about 36 million U.S. jobs by 2035 while the economy generates about 41 million. The harder number is in the middle: roughly 11 million workers may need to shift occupations, about 770,000 transitions a year, which McKinsey puts at 3.6 times the historical average. Only one in seven displaced workers has a direct pathway to a growing job. Roughly 85 percent of growing jobs require credentials or certifications. And in job postings between 2022 and 2026, McKinsey found demand for AI fluency rose about 11-fold, demand for adaptability about fivefold, and demand for willingness to learn, resilience and curiosity roughly tripled.

The line from that report I keep returning to: “Technology determines what is possible, but managerial choices determine how quickly those possibilities become reality.”

Second, the employer view. Fortune’s coverage of the chief people officers at its AIQ Summit (October 1) captured three positions. HPE’s Stacy Dillow said that if organizations are not trying to change how they work, “we’re just trying to retrofit” AI into existing processes. Palo Alto Networks’ Danielle Gonzalez said the company is looking for people who can “not only learn something at a fast rate” but “also unlearn,” and described a move toward observable interviews and hackathons. Lennar’s Drew Holler put it plainly: “We’re going to give you all the tools, but it’s your responsibility to upskill yourself.”

Third, the shop floor. In a Fortune piece on Ford CEO Jim Farley (October 2), Farley, speaking at Ford Accelerate in Detroit, said the line between engineers and hourly skilled-trades workers “is completely blurred out.” The article argues for creating a learning organization for every worker, not just degree holders, and describes Trane Technologies building learning labs for skilled trades workers modeled on athlete training. It also profiles a 17-year-old training as an auto mechanic who sees the trade as one step in a larger plan, not a lifelong box.

What this reveals: unlearning is a leadership task

In The Next Turn, I describe two changes happening at once. The structure of work is changing: how work actually flows, now that AI is becoming infrastructure. The definition of work is changing too: what people believe work is for, what it should provide and what it is worth.

This week’s news sits right where those two turns collide. McKinsey and Ford are describing the structural turn. Jobs are being rebuilt and the boundaries between roles are blurring. The executives are describing a definition: who owns growth when the job keeps moving.

“It’s your responsibility to upskill yourself” is not only a training policy. It is a statement about the exchange. And people evaluate exchanges.

In Chapter 1, “When What Worked Stops Working,” I write about how leaders feel this friction first. Experience becomes less portable. The playbook that built a career stops producing the same results, and it is tempting to personalize what is structural.

What worked is no longer working because the system that made it work is no longer holding in the same way.

The Next Turn, Chapter 1

That applies to leaders as much as anyone. The career ladder you climbed. The signals you used to judge readiness. The assumption that a role stays roughly the same long enough to master. Those are the things the system is no longer holding. Asking employees to unlearn while leaders keep those assumptions intact is not transformation. It is delegation.

Chapter 14, “The Leader as Translator,” makes the point directly. Relevance and growth are workforce requirements, not a Gen Z initiative or a frontline program. Nobody gets to watch this from the sidelines.

Everyone has to move.

The Next Turn, Chapter 14

There is also a warning in that chapter about two errors leaders make in moments like this: romanticizing the past and romanticizing disruption.

Experience without adaptation becomes rigidity, while disruption without judgment becomes recklessness.

The Next Turn, Chapter 14

Unlearning sits between those two. It does not mean throwing out experience. It means keeping what still holds and letting go of what only felt permanent.

Why the exchange matters more than the tools

Chapter 5 is about the value exchange. In an older model, loyalty came first and was tested later. Increasingly, people assess the value first. They ask whether the role offers clear, visible growth.

Now look at McKinsey’s numbers through that lens. If only one in seven displaced workers has a direct path, and most growing jobs require credentials, then reskilling is not a weekend of tutorials. It takes time, structure and a visible pathway. When an organization’s answer is “here are the tools, the rest is on you,” employees hear the terms of the deal. Some will rise to it. Many will quietly conclude the exchange no longer makes sense.

Ford’s blurred line points to something I explore in Chapter 10. Jobs are bundles of tasks, and AI is rearranging the bundles. When the engineer-versus-technician boundary dissolves, the old career map stops working too. The answer is not to tell people to figure out the new map alone. It is to design the system that lets them move. As I argue in that chapter, the work of leadership moves from managing roles toward designing systems.

The standards do not drop. Reliability, quality and sound judgment matter as much as ever. What changes is how people get there, and who is responsible for building the road.

Your next turn

  • Name one thing you need to unlearn. Write down one assumption about careers, readiness or roles that shaped your own path. Ask whether it still holds for the people you lead.
  • Define the learning exchange out loud. Be specific about what the organization provides (time, pathways, coaching, paid credentials) and what is genuinely individual responsibility. Ambiguity here reads as “you’re on your own.”
  • Map the adjacent moves for one role. If this role shrinks or changes, where could the person in it go next? If you cannot answer, you have found a design problem, not a motivation problem.
  • Make learning visible, not just tool usage. Palo Alto’s shift toward observable interviews is a useful prompt. Look for ways to see how people learn and adapt on real work, rather than counting logins.
  • Model it. Tell your team about one way you are changing how you work with AI, including what you got wrong. Unlearning spreads faster when people see leaders doing it.

Frequently asked questions

Who is responsible for AI reskilling, the employer or the employee?

Both, but not equally by default. Employees own their curiosity and effort. Employers own the conditions: time to learn, clear pathways and roles that reward new skills. McKinsey’s report notes that managerial choices determine how quickly AI’s possibilities become reality, which puts leaders squarely in the equation.

What does unlearning mean at work?

Unlearning means letting go of assumptions and habits that worked in an older structure of work but no longer fit. For employees, that might be a familiar workflow. For leaders, it often means old career ladders, fixed role boundaries and signals like tenure or visible effort that no longer reliably show capability.

How many workers will need to change jobs because of AI?

McKinsey Global Institute’s September 2026 report estimates about 11 million U.S. workers may need to shift occupations by 2035, or roughly 770,000 transitions a year. It also finds only one in seven displaced workers has a direct pathway to a growing job, so most transitions will need real support.

The road bends for everyone

The executives at the AIQ Summit are right that the people who thrive will be the ones who can unlearn. I would only add that the same is true for organizations, and for the leaders who run them.

That is the heart of what I wrote about in The Next Turn: leaders now have to say out loud what used to go unsaid. That includes who carries the cost of change.

So before you ask your team to unlearn, a fair question: what are you unlearning right now, and could your team name it?

Sources