Over the holidays, leadership often comes up in conversation. Around dinner tables, on walks, while watching the news. We talk about what good leadership used to look like, what still works, and why so many leaders seem even more tired than they used to.
In our recent piece, The Dismantling of Traditional Leadership, I argued that AI is quietly removing the distance that once protected leaders from execution. This article builds on that idea.
For a long time, leadership followed a clear arc. You worked hard, learned your trade, earned responsibility, and eventually stepped back from the day‑to‑day work. Your job became direction, judgment, and results at a more strategic level.
Most organizations are still built around this idea. They invest heavily in leadership programs that promise a special set of capabilities, as if leadership were something separate from the work rather than embedded in it.
That world is changing. Leadership today feels heavier because the distance between leading and doing the work is disappearing.
Distance made this possible.
Rules, settings, thresholds, and instructions decide how work happens before it even starts. Once those systems run, they repeat decisions automatically, at speed and at scale.
Judgment is no longer distributed over time. It is front‑loaded.
This changes the physics of leadership. There is no cooling‑off period, no gradual correction through conversation. A single choice can instantly affect thousands of outcomes. When something goes wrong, it is visible, traceable, and hard to explain away.
AI does not remove leadership. It removes the distance that once protected leaders from the work itself.
A useful way to understand this shift is to look at pilots.
Pilots operate inside some of the most sophisticated socio‑technical systems we have built. They rely heavily on automation, but they are never allowed to leave the loop. Autopilot reduces load, not responsibility. Pilots are continuously trained to understand how the system works, where it fails, and how to intervene when conditions change. They are also trained extensively in non-technical skills: managing themselves under stress, coordinating the crew, and keeping feedback loops open. This focus exists because most serious incidents are no longer caused solely by technical failure, but by human error amplified by fear, hierarchy, and communication breakdowns.
As pilots gain experience, they do not step away from the cockpit. They fly larger aircraft, longer routes, and more complex airspace. The fundamentals stay the same, but the load increases: more people, more risk, more consequence.
Leadership is moving in the same direction. You cannot lead a system you do not understand well enough to supervise, correct, and take over when automation behaves unexpectedly.
In practice, this now requires three forms of fluency simultaneously.
Systems fluency: understanding how agentic workflows behave, where they fail, and how decisions are translated into repeated action at scale.
Human fluency: understanding how people experience those systems, how stress, fear, and hierarchy affect judgment, and how feedback loops break down if they are not actively maintained.
Cross-domain fluency. Modern systems rarely sit within a single function. Automated workflows cut across product, operations, data, incentives, and customer experience. Decisions made in one area now immediately shape behavior in others. Leaders do not need expert depth in everything, but they do need enough connective understanding to see where handoffs fail and where local optimization creates system-wide damage.
What makes this harder than ever is that these systems and domains are not stable. Unlike pilots, who operate certified aircraft in relatively fixed environments, leaders work in systems that are continuously evolving under competitive pressure, innovation, and changing customer needs. They are not just flying the system; they are reshaping it while it is already in motion.
That makes staying close to the work not optional, but essential.
You are leading whenever your decisions shape outcomes for other people, whether or not you hold formal authority. This is not a soft motivational slogan. It is a structural consequence of how work is done now.
People lead through how they design systems, handle edge cases, explain outcomes, and coordinate across boundaries. Deciding how an automated system treats frustrated customers, when it escalates to a human, or how strict a rule should be may sound technical. But those choices directly shape trust, fairness, and experience.
If you understand the technology but not how people experience it, you build systems that feel cold or unfair. If you understand people but not the system, you create rules that fail silently and repeatedly. In both cases, the mistake is built in and repeated at scale.
The abilities required to do good work are the same as those required to lead: domain understanding, focus, judgment, responsibility, and the ability to manage yourself and others well. What changes with leadership is the load: more scope, more pressure, more accountability. That is also why leadership still deserves greater reward. The cost of being wrong is higher.
Most organizations will have to upgrade their approach. They send leaders to off‑sites to talk about vision while critical decisions are being locked into systems back home. They promote people without exposing them to the workflows they now shape. They separate technical training from leadership training, as if judgment about systems and judgment about people could be learned independently.
The result is predictable. Leaders approve changes they do not fully understand. Teams absorb the fallout in real time. Workarounds emerge to protect customers from the system rather than improve it. Leadership looks present on paper, but absent where consequences unfold.
This also changes cross‑industry leadership. In the past, leaders could rely on general management skills while learning the domain over time. Today, shallow understanding shows up immediately when decisions are executed at speed. Cross‑industry leadership is still possible, but slower, riskier, and more demanding.
Hierarchy still exists. It remains the least-bad way to decide who bears final responsibility. But hierarchy no longer defines who actually leads.
This changes how leaders must be developed. Abstract cases and detached theory are no longer enough. Leaders need training that keeps them inside real systems while increasing load: exposure to end‑to‑end consequences, practice under pressure, and simultaneous development of technical and human judgment.
The leaders who will do well are not those who rise above the work, but those who can stay close to it while carrying greater responsibility. They understand what the system does, how people experience the results, and they are willing to own the outcomes.
If you are still building separate leadership frameworks, ask yourself a harder question: are you solving a real problem, or defending a distinction that AI has already dissolved?