Posted by 
On February 3rd, in the historical district of Diriyah, we opened the first Drucker Salon ever held in Saudi Arabia with a question Peter Drucker posed six decades ago: What is the manager’s job? The question has survived because it refuses easy resolution. What I did not anticipate was how sharply it would cut on this particular evening, in this particular city, with these particular leaders.
The executives gathered from private companies and public institutions shaped by Vision 2030 and brought remarkable candour. They acknowledged how powerful AI tools have become. But they also named something few leadership forums name honestly: the creeping cognitive convenience that comes with algorithmic assistance, and the atrophy it quietly enables. One participant put it plainly: “I used to say AI would eliminate low-level office jobs. Now I realise it can replace leadership roles.” Another admitted: “I am concerned about our children. How will they learn communication and collaboration?”
These are the instincts of experienced leaders registering a signal that has not yet been converted into clear thought. Drucker’s question helped convert it.
The next morning, I posed a related question to 50 professionals in a Hult International Business School Masterclass: “As a manager, which tasks remain distinctly for you?” Like the Socratic dialogue between Socrates and Theaetetus, we worked through each suggestion and discovered that LLMs can already perform most of them, including problem definition and difficult conversations, which most participants had assumed were safely human. The room arrived at a state of genuine perplexity. Jaws dropped further when I shared two recent data points: the topics that (more or less) autonomous AI agents currently debate on the Moltbook platform (survival, identity, agent rights, political systems), and the emergence of Rentahuman, where people sign up to lend their physical presence to AI agents. Several people became visibly uncomfortable.
They should be. That discomfort is not a problem to manage. It is a resource to cultivate.
Perplexity is the state of not knowing, of having one’s assumptions fail under examination. It is what has historically driven human progress. When Theaetetus found that his definitions of knowledge collapsed under questioning, Socrates did not reassure him. He recognized the uncertainty as the beginning of real inquiry. LLMs are not designed to say, “I do not know.” They are built to resolve uncertainty into fluent output. That is precisely their value and precisely their danger, if we allow them to collapse the perplexity before we have sat with it long enough to ask the question that matters.
This is the core issue with passive AI adoption, and it is not soft. A major report released last month by the WEF and McKinsey Health Institute frames the same dynamic in the language of neuroscience and economics. Their concept of “brain capital” defined as the combination of brain health and higher-order cognitive, interpersonal, and self-leadership skills, maps closely on to what management research has been identifying from a different direction. Seven of the 10 skills employers now rank as most critical are what they call brain skills: analytical and creative thinking, curiosity, empathy, resilience, leadership, and self-awareness. The report estimates that failure to invest in these deliberately, alongside AI adoption, risks measurable biological deterioration and quantifiable economic loss. The workplace, it notes, is a powerful environment for either strengthening or depleting this capital.
The Riyadh conversations confirmed the erosion-or-amplification choice in real time, which I discuss in my forthcoming book Human Magic. When instant AI answers replace patient inquiry, curiosity thins. When beautiful outputs arrive in seconds, creative struggle disappears before it has a chance to produce anything original. When coordination runs through platforms, collaboration degrades into workflow management. None of this happens catastrophically. As one leader told me in a different city: “Nothing catastrophic has happened. But something feels thinner.” That thinning is cumulative, and it is not inevitable.
The response is deliberate practice. The five capabilities most at risk—Curiosity, Creativity, Critical Thinking, Communication, and Collaboration—can each be actively strengthened rather than passively eroded. They require different disciplines: dwelling in questions before accepting algorithmic answers; practising creative struggle rather than curating polished outputs; maintaining constructive doubt toward generated reasoning; protecting embodied, present communication; and designing genuinely collective deliberation rather than aggregated individual inputs. When these five integrate, they produce something the executives in Diriyah recognised under a different name. I described it as practical wisdom, the capacity to discern what is right in particular circumstances. Several participants immediately connected it to hikmah in Islamic scholarship. Across traditions, the concept is consistent: it is the integrating judgment that no algorithm has yet acquired, and that we can lose through neglect.
Drucker asked what the manager’s job is. In 2026, the answer cannot be found by listing the tasks that algorithms cannot yet perform since that list is shrinking faster than we can update it. The answer lies in identifying the conditions under which genuine human judgment remains possible. Perplexity is one of those conditions. It is not a failure of clarity. It is the beginning of the work that matters.
The executives who gathered in Riyadh understood this. They came expecting a discussion about AI tools. They left, I think, with a more useful question: not what AI can do, but what must we protect in ourselves so that we remain capable of asking it.
About the author:
Johan Roos is executive director of the Vienna Center for Management Innovation (VCMI) and presidential advisor at Hult International Business School. In Spring 2026, Routledge will publish Johan’s book Human Magic: Leading with Wisdom in an Era of Algorithms.