14 July 2026 · René de Baaij
English Leadership & AIHuman-AI
AI Leadership: Why the Black Box Calls for a Different Kind of Leadership
In most organizations we lump leadership and management together. That is not even unreasonable. We expect managers to show leadership, and we expect them to carry out management tasks. The two are rarely separate. But they are different things, and right now the difference has started to matter.
Management is about arranging and organizing: allocating people and resources, defining objectives, making sure a group delivers a result together. That is the process in the surface current: the rational, the factual. Leadership is about something else: the meaning we give to that work. People want to be part of a community, to lead a meaningful life, to do meaningful work. That lives in the undercurrent, and in most organizations it gets too little attention.
As long as eighty or ninety per cent of a manager's energy goes into management, it is worth asking yourself: where is the leadership in this issue, or am I only managing? A rule of thumb that helps me: a manager gives answers, a leader asks questions.
Why the old paradigm worked for so long
We have built a management culture in which a few things are sacred: cost control, predictability, repeatable results. We try to create the simplest, most effective process possible, one that delivers the same output over and over. That is a convergent process. We keep improving, but within the objective that is already in place. With it comes the fact that we do not say to people: be creative, come up with something new. We say: do the same again tomorrow, controlled and consistent. If everyone does that, you have an efficient machine. Management is happy, because it can be measured. Spreadsheets, KPIs, and performance looks splendid.
So what is the problem? Even without AI, we already see people with more to offer walking away. Creative people who drop out after a few years in a management role because they could do more but are given no room for it. People who want to contribute to something of social value and do not feel that in their organization. None of this is new; it has been described for a very long time. As long as the machine runs and makes money, everyone is content.
But large-scale change lies ahead. Markets can collapse, oil prices fluctuate, geopolitical relations shift and feed through into exchange rates. An organization has to learn to deal with that. And then along comes something that moves at breathtaking speed: AI. It multiplies the problem a hundred or a thousand times. We are not moving from evolution with normal fluctuations to revolution; we will see companies disappear because someone else does something entirely new with AI. In software development the first reports are already in: work that took a team a year now takes three weeks.
That sounds like a splendid development for a machine that runs well. Except that realizing it requires very different skills. We work at a different level of abstraction, predictability declines, we have to be far more experimental and place far more trust in people's creativity and in their ability to reason backwards: what did we put into that black box, and how does it come out? Those are exactly the routines we have stripped out of the existing system. We are moving from convergent to divergent, from simple and repeatable to complex. And AI acts as an accelerator. Where you once had weeks to experiment at leisure, you now have an answer after three minutes that you have to make sense of.
That calls for different management, and different leadership. That is why I call it AI leadership. Not leadership by AI, not being led by AI, but the question of how leadership must develop in order to deal well with AI.
Back to learning how to learn
In theory the answer is not even that complicated. We need to move from an attitude that is always geared to performance towards an environment for development. We need to learn how to learn again, and as leaders we need to hold the conversation far more from meaning. Part of the ordering and managing of work will move into the systems themselves. The professionals who work with those systems will therefore want to be led differently as well.
There has long been a distinction between first-, second- and third-order learning. First-order learning is improving within what you are used to. Second-order learning asks whether you can improve the processes themselves: letting go of paradigms and looking differently, as we know from business process redesign. But even that does not answer what AI raises. We are now at the level of meaning. What are you still, as a professional, once we introduce this? What is the role of your department if everyone can do this? What is our place in society, if society itself is shifting? Those are the questions leadership has to open a dialogue about. I do not believe anyone has the answer. I do believe it is necessary to have the conversation and find a way forward together.
Twenty years ago there was already a great deal to learn about third-order learning and processes of transformation. Most managers put those books aside long ago and became very good at the other part: first-order learning, improving what already exists. Now AI is washing over us, and all the principles of that older third-order learning need to come back to the table.
AI literacy is not the same as AI awareness
At the end of 2025, Microsoft and Google deliberately chose to make their AI widely available within existing systems. Suddenly everyone had Copilot on their screen, and anyone could download ChatGPT. The result: almost everyone now uses it, from my ninety-one-year-old father-in-law to board members who let Copilot sort their morning email without asking how they will make sure the one important message does not slip through.
As a society, as business and as government, we are learning that this is a perfectly ordinary tool. It is not. Everything you put into it is stored. Everything you do with it carries an inherent inaccuracy, because it is a probability predictor. If we carry that into processes where we are used to putting people to work, we will organize those processes entirely differently. That has major consequences for our existence, our society, how we earn a living and how we matter. Perhaps it will not go as fast as everyone thinks. But it is inevitable. I know of no technology that humanity has ever invented and then abandoned.
What we mostly steer on today is AI literacy: courses on working better with Copilot. The real question is AI awareness. Are you aware of what is coming at us, and what it does to people on the shop floor? I hear plenty of stories about employees who use their own chatbot on their private phone to solve work problems. That calls for a dialogue: what risk are we running here, what does it mean in the long term, and what do you do with people who realize that their work may be redundant in three years? That is an existential question, and in my view it belongs to leadership. The average manager thinks: I am the director here, so people come to me with questions. Leadership takes the nerve to say: I don't know either, but I want to get through this together with you. That is a different language from the one we grew up with. Perhaps we are capable of it. We simply have not practised it for a long time.
Why the conversation does not happen
I meet many people who are racking their brains over this. I meet few who actually take it on. Especially in the information domain I often hear: I am still straightening out the implementation from five years ago; once that is done, I will think about AI. But that is exactly the problem. It is as if nothing is changing, while the fundamental question is left untouched. Do we need entirely different strategies, and what does this mean for people in their work?
Beneath this lies a dangerous pattern. We already see it in the United States, in the reports from software development: where do young people still learn, if the whole work process sits inside the black box? They can put something in and take something out, but they do not understand how a result came about when the outcome is wrong. So the question is how we pass on experience to young people in a way that lets them work with AI while keeping the knowledge and logic to judge whether things are going well.
I saw this quite literally at a large organization where a steering group had been set up for the introduction of AI. I was sitting with the HR director and asked: what now? She said: I wasn't invited. Apparently it was not about people, then. She had to fight her own way in. That is precisely the evidence that we have no notion of the impact and that we are not putting the right people at the table.
There is a role here for information managers and information professionals too. Because it concerns AI, everyone looks to IT. We used to ask them to reset a password; now they are suddenly expected to lead the conversation about AI. I think senior IT management should take up that role: engage their peers at board level, going further than tightening privacy rules or sending an email about banned apps. The fundamental conversation: what is coming at us, and how do we become better together without drowning in it?
The average IT professional will agree with that, and at the same time say: we are a voice crying in the wilderness. We have been warning for ten years that AI makes mistakes and that you should not feed data into a system just like that, and nobody listens any more, because it is so easy. That wagging finger is exactly where it goes wrong. What does work is opening the dialogue: I don't know exactly what is happening to us either, but neither do you, and we have something to do together. That is a conversation at a different level: the leadership conversation.
My role in this is not new; I have been doing this work for a long time: getting that conversation started, facilitating it, guiding it, and making sure it happens in the right way. When a CIO tells me that nobody listens to him, the first thing we need to repair is his position, so that his message is taken seriously. Usually something has happened over the years that turned him into "that person from that department" who never delivers anyway. Let them get the existing implementation in order first, the thinking goes, while those screens may no longer be relevant in two years.
None of this comes from unwillingness or incompetence. People are much smarter than what they use, and capable of much more than what they apply. It lies in the patterns of the organization, which carry implicit limitations. Those patterns need to be broken, and that takes people who dare to ask an awkward question in the undercurrent now and then. Not by piling up knowledge, but through experiential learning: putting people in a setting where they talk with one another, without immediately reducing it to "so this is what you always do". Because once someone is constantly disqualified, he stops worrying about whether AI will eat the company in three years.
What you can do tomorrow
You do not need to hire an adviser straight away. Simply ask your people: does this keep you awake at night? You read the papers too: what is happening to us? Are we having the right conversation in the boardroom, about the things that truly concern our existence in the long run?
I know an organization whose greatest problem is retaining young people. Since everyone started working from home en masse, staff come to the office only one or two days a week. Young people find it dull, learn nothing, make no connections, and are gone again after four months, while the profession this organization practises takes years to master. Add AI to that and the problem only grows: in five years this organization may no longer be able to do its work, while it manages billions. That is existential. It calls for a discussion about what needs to be organized so that people learn the profession, are able to practise it with AI in five years, and want to commit to it with their heart and their career. Everyone agrees with me on that. And everyone also says: we are so busy. We know that excuse. It only gets busier if you do not go through this.
Most organizations remain stuck in first-order learning, with the occasional excursion into the second. The step to the third does not come, simply because it requires an investment that cannot be quantified. You know it matters, but nobody can tell you what it will deliver in concrete terms two years from now. In a shareholder culture, and in government the same goes for the political culture, short-term results weigh more heavily than long-term performance. We simply do not allow ourselves the calm and the space to sit back and think about what is coming.
Some organizations already do something about it, only in the wrong way: one away day a year, a bit of reflection, letting off steam, and afterwards nothing changes. What I believe in is institutionalizing it. Simply say it out loud: every fortnight we have a day on which we do nothing but think about this. At first that feels like a loss. But what I see in practice is that once you build in that space deliberately, people turn out to get the same amount of work done in one day less, or they consciously let certain things go. Bring your people along in this and they will take responsibility for it. Performance stays on track. Only the conversation that comes before it is one we first have to learn to hold again.