The more leadership is mediated through technology, the greater the risk is not of loss of control but of a gradual loss of proximity.
During a recent travel disruption caused by a Lufthansa strike that turned what should have been a short flight into a 13-hour journey, I found myself with something increasingly rare: uninterrupted time to read a printed newspaper and reflect. Among the articles I went through in the Financial Times, one stood out—not because it introduced a radically new idea, but because it articulated a question that many CEOs are already confronting, whether consciously or not: how much thinking should we delegate to AI?
The article referenced Mark Zuckerberg and his exploration of AI agents designed to extend his presence as a leader—systems capable of communicating, gathering information, and potentially acting on his behalf across the organization. At first glance, the concept is compelling. The idea of leadership that can scale almost infinitely, supported by systems that never tire, never lose focus, and can process complexity at extraordinary speed, speaks directly to the increasing demands placed on executives today. And yet, the more one reflects on it, the clearer it becomes that this is not primarily a technology discussion. It is a leadership question in its purest form.
There is no denying that AI fundamentally shifts what is possible at the executive level. Leaders can now access synthesized insights across functions in real time, identify patterns that would otherwise remain hidden, and reduce the noise that so often clouds judgment. In an environment where speed, precision, and clarity are critical, these capabilities are not simply helpful; they are transformative. Used thoughtfully, AI allows CEOs to focus less on processing information and more on interpreting it, less on reacting and more on deciding with intent.
However, this is precisely where a subtle but important tension begins to emerge. Efficiency, no matter how powerful, is not the same as leadership. Organizations are not purely rational systems that respond predictably to optimized inputs. They are shaped by human dynamics—by trust, informal structures, unspoken concerns, and the emotional undercurrents that rarely surface in formal reporting. Even the most advanced system can only approximate these realities, and often it cannot see them at all.
This distinction matters because there is a growing temptation to equate better data with better leadership. When information becomes cleaner, faster, and more accessible, it creates the impression that decisions themselves will naturally improve. But data, no matter how refined, does not carry the same context as human experience. A recommendation can be analytically sound and still be strategically flawed if it fails to account for the realities beneath the surface—realities that are often intangible, difficult to measure.
It is here that an older leadership principle, often dismissed as outdated, becomes highly relevant again. The idea of “management by walking around” may sound almost nostalgic in a world increasingly defined by digital interfaces and remote interactions. Yet, its underlying logic is more important than ever. Leaders who step beyond dashboards and reports and engage directly with their organizations gain access to something that no system can replicate: the texture of reality. They notice hesitation where reports show alignment, they sense shifts in energy that no metric captures, and they hear concerns that would never formally be raised.
These moments are not incidental; they are central to effective leadership. AI can identify anomalies, detect patterns, and even flag potential risks. Still, it cannot interpret silence in a meeting, nor can it fully understand the meaning behind a delayed response or a carefully chosen phrase. Presence, in this context, is not a matter of personal style; it is a mechanism of governance. It builds credibility, and credibility, over time, is what enables alignment across an organization.
As leadership becomes increasingly mediated through technology, the risk is not that CEOs will lose control, but that they will gradually lose proximity. And proximity, while often underestimated, is critical for sound decision-making. When leaders rely too heavily on filtered insights, they begin to operate at a level of abstraction that can distance them from the lived realities of their business. Decisions can then be based on representations of reality rather than reality itself, which is where misalignment often begins.
There is also an accountability dimension that cannot be overlooked. As AI systems become more integrated into analysis, communication, and decision support, the question of responsibility becomes more complex, not less. AI can inform decisions, but it cannot own them. It does not carry accountability for outcomes, nor does it bear the consequences of flawed judgment. That responsibility remains firmly with leadership.
This introduces a new expectation for CEOs, one that extends beyond simply using technology effectively. Leaders must act as stewards of the systems they rely on, ensuring not only that they are efficient but also appropriate. This means questioning outputs that appear correct but may be strategically misaligned, recognizing potential biases embedded in data, and ensuring that decisions remain consistent with the organization’s values and long-term direction. In this sense, the role of the CEO is not diminished by AI; it becomes more demanding.
Framing the relationship between AI and leadership as a competition misses the point entirely. The more relevant perspective is one of augmentation. AI should extend leadership’s reach, enhance its clarity, and free up capacity for deeper thinking, but it should not replace the core elements that define it. The most effective leaders will be those who use AI to sharpen their judgment while remaining firmly anchored in the human realities of their organizations.
The more subtle risk, and perhaps the more dangerous one, is not that AI will take over leadership in any dramatic sense, but that leadership itself becomes increasingly abstract. When every interaction is mediated, every signal filtered, and every decision pre-processed, leadership can appear highly efficient while gradually losing its connection to what is actually happening within the organization. And once that connection weakens, performance inevitably follows.
For CEOs, the challenge is therefore not whether to adopt AI, but how to integrate it without losing what makes leadership effective in the first place. This requires a conscious balance—leveraging technology to enhance insight while maintaining direct engagement, delegating tasks without relinquishing responsibility, and embracing efficiency without sacrificing presence.
Because ultimately, leadership still depends on what cannot be automated: the ability to understand context in its full complexity, to build and sustain trust, to exercise judgment in uncertain situations, and to take responsibility when outcomes are unclear or uncomfortable. No system, regardless of its sophistication, can fully replicate these qualities. And that is precisely why they remain at the heart of the CEO’s role.