For the past decade, companies lowered their investments in the experiences that build leadership judgment: stretch roles, cross-functional assignments, and real ownership of results. They did it to save money, but kept the same leadership development model built for stable, layered organizations.
That world doesn’t exist anymore. Since Covid, organizations have gotten flatter and less predictable, and AI has changed how daily work gets done.
The result: leadership benches are thinner than boards expected. Talented people are there, but they haven’t had the experience that used to turn talented people into senior leaders.
AI strains this in two ways. It raises the bar for judgment, because someone still must check whether its output is right. And it takes over the hands-on work that used to teach people that judgment in the first place.
We call this gap the development debt: skipped experience that companies need later, and eventually have to pay for. Our new playbook, Getting Ahead of the Leadership Gap, co-published with Charter, maps this debt and how to start closing it.
Diane Gherson, IBM’s former CHRO, outlined the stakes in a single question in her interview for the playbook: “The next generation of managers won’t have done the work… how on earth are they going to manage these agents?”
The playbook compiles the numbers behind the strain:
Years of cost constraints removed the middle layers where rising leaders learned to lead. People in those roles had cross-functional exposure, real accountability, and a broader network of relationships. Those roles intentionally prepared people for senior roles; with fewer opportunities there was less chance for apprenticeship.
The scale of the flattening is measurable: Gallup data cited in the playbook shows the average team size grew roughly 50% between 2013 and 2025, to just over 12 people per manager. Wharton’s Peter Cappelli, interviewed in the playbook, traces the slide back to 1981, when companies began unbundling the “great corporation” model of rotational assignments and ten-year succession plans: “Squeezing headcount, cutting out middle management, expanding spans of control—all that stuff made things noticeably weaker.”
The incentives hold the pattern in place. In the same playbook interview, Cappelli points to the two performance metrics investors watch most, profit per employee and revenue per employee: “If you cut headcount, you cut the denominator… That’s driving a lot of this.” The leaders making those calls have changed too. “You get a lot of engineers who are now CEOs and when they think about employment, they think about optimization,” he says. In his telling, no one is rewarded for developing the organization from within, and development is the first budget to go. He sees it show up in three ways:
Boards today need leadership capabilities the old model rarely built, and those take the longest to grow. The playbook calls them the three catalysts, sitting alongside the enduring essentials of business acumen, individual coaching, and technical literacy:
Gherson’s advice in the playbook is that the discipline that matters most is knowing where the line sits. “That’s always been my motto with AI: just because it can doesn’t mean we should.” Deciding when to use AI and when to leave the work to humans is, in her words, “the future of leadership development.”
Leaders need more domain depth to judge AI’s output Matt Beane, associate professor of technology management at UC Santa Barbara and author of The Skill Code, named the strain in his playbook interview: “Quality assurance is the bugbear of the AI economy,” he says. As AI multiplies the volume of work a single manager must vet, “people with more expertise are going to feel more and more… [of a] burden for deciding whether something’s shippable or not.”
The same risk runs forward in time. If junior employees do less of the hands-on work that builds expertise, reach the leadership level without having done the work themselves. An Emory University experiment by professor William Schmidt, cited in the playbook as the “promotion paradox,” found exactly that pattern: participants who used AI got “promoted” through early tasks much faster, then completed fewer of the harder tasks correctly once AI was taken away. “They were using AI to speed through their pre-promotion phase,” Schmidt said, “without having struggled through the process that’s necessary to really internalize how to do those problems.”
A thin pipeline makes every other gap worse. Fewer leaders means wider spans for those who remain, wider spans leave no time to coach the people coming up, and AI raises the bar on the judgment those under-coached people will soon be expected to show. In the playbook, Kevin Cox, the former CHRO of General Electric, American Express, and Pepsi Bottling Group, describes this issue clearly: “I feel like we’ve made what used to be jumps over a river like jumps over the Grand Canyon. AI is potentially going to turbocharge that.”
Spending more on the same development model leaves the debt in place. The companies making progress start from a different question: which experiences did we cut, and how do we rebuild them deliberately?
The expertise still sits with today’s leaders, and that is the trap: the window closes as they move on. The companies moving first are rebuilding the missing experiences now, before they promote a generation of leaders who never had them.
Read the full playbook: Getting Ahead of the Leadership Gap (Council Advisors × Charter), or download the PDF from Charter.
What is the development debt?
Companies spend years cutting the experiences that build leadership judgment, like stretch roles and cross-functional assignments, then need that judgment anyway. That gap is development debt. Like financial debt, it builds quietly and comes due all at once.
Why does AI make the leadership gap worse?
AI takes over the hands-on work that used to build expertise. But someone still has to judge whether its output is right, and that judgment takes exactly the expertise the work used to build. AI removes the training ground and raises the bar at the same time.
What should companies do first?
Audit leaders against the three catalysts in the playbook (human skills, cross-functional capability, and AI fluency) to find the widest gap. Then rebuild the specific experiences that got cut, instead of adding new programs to the old model.
We use data collected by cookies to analyze traffic on this website. By clicking “Yes,” you agree to our use of cookies as described in our Privacy Policy.