13 October 2026 · René de Baaij
Essay for webinar 1.2, "Relevant perspectives on leadership and AI", Tuesday 27 October 2026. Reading time about twenty minutes.
Three perspectives on leadership in times of AI
The essay Who are you when you have no one to convince?, which underpins Cycle 1, starts with self-knowledge. A leader who does not know why he does what he does leans on habit and on what others expect of him. This essay takes the next step. It asks what your view of leadership rests on. To answer that, it sets three perspectives from leadership theory side by side. Each places leadership somewhere different: in the position, in the person or between people. That location determines how vulnerable a view of leadership becomes when technology takes over part of the work.
The occasion is the arrival of AI in the everyday work of organisations. The economists Ajay Agrawal, Joshua Gans and Avi Goldfarb (2018) describe AI as a prediction technology. A system learns from past data and estimates what is likely: which customer will cancel, which applicant will fit, which schedule is least likely to break down. Their central claim is economic. When prediction becomes cheap, the value of what is needed alongside it rises. That is above all judgement: deciding what an outcome is worth, which error is worse than another, and what to do with a prediction that is accurate but not right. In a later book the same authors show that the largest changes arise when organisations redesign their systems around that cheap prediction. At that point, who decides shifts as well (Agrawal et al., 2022).
The philosopher Brian Cantwell Smith (2019) draws a comparable distinction. He calls the calculating and estimating of systems reckoning and contrasts it with judgment: the kind of judgement that knows what it is about, feels committed to reality and takes responsibility for what it concludes. AI is strong at the first. According to Smith, the second has to be deliberately designed, protected and practised.
This puts three questions on the table that run through the whole essay. Who may decide? On what basis? And what remains human work? Each of the three perspectives answers those questions differently.
1. Leadership as position and function
The oldest perspective is also the most familiar. Here leadership sits in the structure of the organisation. Whoever leads does so because he or she holds a formal position. The German sociologist Max Weber (1922) distinguished three sources of legitimate authority: tradition, charisma and legal-rational authority. The last type rests on rules and procedures. People obey the office rather than the person. A director has authority because the appointment followed the rules, and that authority passes to the successor. For Weber this was the core of modern bureaucracy. It makes organisations predictable, protects against arbitrariness and makes cooperation on a large scale possible.
The French engineer and executive Henri Fayol (1916) gave this perspective practical form. He described the work of the manager as five functions: forecasting and planning, organising, commanding, coordinating and controlling. A century later, that division still shapes job profiles, management training and annual plans. Whoever leads manages resources, allocates work and oversees its execution.
This perspective deserves more appreciation than it often receives in current leadership thinking. It makes clear who is responsible and to whom you can turn. It gives employees certainty about what is expected of them and what they can count on. Weber also saw the downside. A society that runs entirely on rational rules can become a shell from which no one can escape. He used an image for this that later became known as the iron cage (Weber, 1905).
For the question of AI, this is the perspective that is hit most directly. Much of what Fayol described is information work. Planning means predicting how much work is coming and how many people it will take. Organising and coordinating means agreeing who does what, based on what is known. Controlling means comparing what happens with what was expected. These are exactly the tasks at which predictive systems excel. Rosters, capacity planning, inventory management and performance dashboards have for years been increasingly prepared or carried out by systems.
Shoshana Zuboff (1988) saw this early on. In her research on the introduction of computers in factories and offices, she distinguished two possibilities of information technology. Technology can automate work, and it can informate work: it makes visible what was previously hidden, to anyone with access to the data. That second possibility undermines the position of the manager. Whoever used to draw authority from an information advantage sees that advantage disappear once everyone has the same dashboard.
There is a less visible risk as well. The position can remain while its substance disappears. Madeleine Clare Elish (2019) studied what happens when people and machines together form a system. She found that the human in such systems often serves as a moral crumple zone. When things go wrong, responsibility is placed on the human operator, even when that person had little influence over what the system did. For managers this is a real situation. You sign off on a schedule, a selection or an assessment prepared by a system. Formally, you are the one who decides. In practice, you follow a proposal whose basis you cannot verify.
The position perspective therefore gives a clear answer to the question of who decides: the person in the position. The answer to the question of on what basis is increasingly taken out of their hands by AI. What remains is the judgement that Agrawal and Smith point to. A manager who draws authority from the position has to ask whether he still truly exercises that judgement, or whether he mainly signs.
Reflection question 1. What part of your work as a manager consists of planning, allocating and controlling? And what would your authority rest on if a system took that part over from you?
2. Leadership as a personal quality others see in you
The second perspective moves leadership from the position to the person. Someone leads because he or she has certain qualities: vision, self-confidence, persuasiveness, decisiveness. For a long time, leadership research searched for the traits that distinguish leaders from non-leaders. That search yielded less than expected. In an influential review, Ralph Stogdill (1948) concluded that traits predict little unless the situation is taken into account. Someone who stands out as a leader in one context does not do so in another.
Even so, the image of the leader as an exceptional person never went away. Here too, Weber (1922) laid the foundation. Alongside legal-rational authority he described charismatic authority: authority that rests on extraordinary qualities that followers see in a person. Weber emphasised that recognition. Charisma exists for as long as followers grant it, and it must prove itself again and again. If success fails to come, the authority disappears with it.
Later research developed this idea further. Robert Lord and Karen Maher (1991) showed that everyone carries an implicit image of how a leader looks and behaves. They call that image a prototype. We perceive someone as a leader when that person fits our prototype, and we infer leadership from good results. In this approach, leadership is to an important extent a matter of perception. James Meindl, Sanford Ehrlich and Janet Dukerich (1985) went a step further. In a series of studies they found that people use leadership as an explanation above all when an organisation's performance is exceptionally good or bad. They called this the romance of leadership: we attribute the outcomes of complex organisations too readily to the person at the top.
Manfred Kets de Vries (2001) adds a psychodynamic layer. In their relationship with a manager, employees bring along earlier experiences of authority, often from childhood. They project expectations, desires and fears onto the leader. That explains why leaders sometimes evoke more admiration or resistance than their behaviour warrants. Popper and Mayseless (2003) show in a similar way that the relationship between leader and follower resembles the attachment relationship between parent and child. In this light, charisma is partly a projection. It lies as much in the eye of the follower as in the person of the leader.
For the question of AI, this perspective is instructive in an unexpected way. People project onto systems too. A language model that writes fluently, assertively and without grammatical errors quickly creates an impression of expertise. That impression is not always justified. Ethan Mollick (2024) describes how AI has a jagged frontier of capability. The system performs excellently on some difficult tasks and fails at some simple ones, without it being visible from the outside where that frontier lies. The prototype of the expert, who speaks fluently and never doubts, is served better by a system than by most people.
That has two consequences for managers. The first is competition in the field of persuasiveness. Whoever draws authority from the image of the expert who knows everything sees that image fade beside a system that answers any question in seconds. The second consequence is subtler. Managers may start borrowing the authority of the system. "The model indicates that …" sounds more objective than "I think that …". The source of authority then shifts from the person to the system, while the person remains formally responsible. That brings us back to Elish's moral crumple zone.
The person perspective answers the question of who decides with: the one whom others recognise as a leader. The question of on what basis gets an uncomfortable answer: on the basis of what followers see in someone, and that image sometimes says more about the follower than about the leader. What, then, remains human work? Above all, the ability to see through that image, in yourself and in others. That connects with the self-knowledge with which Cycle 1 began.
Reflection question 2. Which qualities do others attribute to you as a leader, and how much of that stems from their expectations? What remains of that image when a system formulates more confidently and faster than you do?
3. Leadership as a meaning-making pattern between people
The third perspective places leadership in the interaction between people. Here leadership is a process in which several people take part and in which roles can change. Linda Smircich and Gareth Morgan (1982) gave this perspective a sharp formulation. They described leadership as the management of meaning. Every organisation is constantly confronted with situations that can be understood in more than one way. The leader is the one who manages to name the situation for others, so that they know what is going on and what is being asked of them. Smircich and Morgan also pointed to the downside. Whoever defines reality for others makes them dependent on that definition.
Karl Weick (1995) worked out the underlying question under the heading of sensemaking. People in organisations act first and understand afterwards what they have done. Together they look for a story plausible enough to carry on. Meaning arises in conversation, in the choice of words and in what is and is not named. Whoever influences that influences what the organisation does.
Alexander Haslam, Stephen Reicher and Michael Platow (2011) linked this perspective to social identity theory. They argue that leaders have influence to the extent that they manage to embody and shape a shared "we". They distinguish four aspects. A leader must be recognisable as one of us. He must stand up for the group. He must help define who we are and what we stand for. And he must make that "we" visible in structures and practices, so that it matters. In this approach, leadership is the work a group does on its own identity, with the leader as the person who carries that work most visibly.
Mary Uhl-Bien (2006) makes the distinction between this perspective and the previous two explicit. She contrasts an approach that sees leadership as a property of a person with a relational approach, in which leadership emerges in the social processes between people. In her later work with Russ Marion on complexity leadership, that idea is extended to organisations as a whole (Uhl-Bien & Marion, 2009). Ralph Stacey (2001) describes organisations in comparable terms, as patterns of conversation that continually renew themselves.
For the question of AI, this perspective offers the most to hold on to, but it is no free pass. At first sight, meaning-making seems to be the domain that systems cannot reach. A system can predict what is likely. It cannot determine what something means to a group, and it is not part of the "we" that is looking for direction. In Smith's terms, the system lacks the commitment to reality that makes judgement possible.
Wanda Orlikowski (2007), however, warns against too strict a separation between people and technology. She shows that technology becomes part of the practices in which people work together. That applies to meaning-making as well. Whoever lets an AI system summarise the minutes lets the system decide what was important. Whoever has an analysis drafted lets the system choose the concepts in which a problem is described. The system thereby takes part in the management of meaning, without anyone having decided so and without the system being able to bear responsibility for the meaning it offers.
The interaction perspective therefore answers the question of who decides differently from the previous two: everyone who takes part in the conversation in which the situation is named, and that can include a system. The question of on what basis becomes: on the basis of the meaning that emerges in that conversation. What remains human work is guarding and carrying that meaning. That requires a leader to notice when the words of a system become the words of the group, and to help the group decide for itself what something means. A system can predict, but it cannot want. The wanting, and the responsibility that comes with it, stays with the people who look for direction together.
Reflection question 3. When did you help others understand what was going on in the past month? What exactly did you do at that moment, and which part of it could a system have done?
4. Three layers in one practice
The three perspectives are often presented as a progression from outdated to modern. That does them an injustice. In practice they operate at the same time. Every manager holds a position with formal responsibilities. Every manager is perceived by others and has qualities attributed to them. And every manager takes part in the conversations in which meaning arises. The question is which layer weighs most heavily in how you understand your own leadership.
AI makes that question more urgent, because the technology affects each layer in a different way. The position perspective loses part of its substance, because planning and controlling can be done by systems. The person perspective faces competition, because systems serve the image of the fluent, confident expert better than people do. The interaction perspective remains the least replaceable, but it has to be maintained deliberately, because systems take part in the conversation unnoticed.
Agrawal, Gans and Goldfarb predict that judgement will become more valuable as prediction becomes cheaper. This essay adds a question. Where does that judgement sit in your practice? In the position perspective, it lies in the responsibility you take for a decision. In the person perspective, it lies in the ability to see through what others project onto you and onto systems. In the interaction perspective, it lies in caring for the meaning a group makes together. A leader who leans only on the first layer runs the greatest risk of having his role hollowed out. A leader who draws authority mainly from the second layer is challenged most by the fluency of systems. A leader who takes the third layer seriously has the most work to do, because that is precisely where the human contribution cannot be taken for granted.
5. Back to self-knowledge
The essay for Cycle 1 asked who you are when you have no one to convince. These three perspectives make that question more concrete. Whoever draws authority mainly from position has no one to convince as long as the structure holds. Whoever draws authority from the person is constantly busy convincing, even without experiencing it that way. Whoever understands leadership as shared meaning-making convinces less and listens more, but must be willing to take a position in the conversation.
So the point is less which perspective you find most attractive. The point is which perspective you actually use, and why. Do you recognise yourself in the position perspective because it works, or because it gives you certainty? Does the person perspective appeal to you because you have the qualities, or because it is pleasant to have them attributed to you? Do you choose the interaction perspective because you give direction through conversation, or because it relieves you of the duty to decide yourself?
AI makes these questions more pressing, because technology takes over, without hesitation, the parts of leadership that depend least on who you are. What remains does depend on you. In webinar 1.2 we explore together how the three perspectives relate to one another in your own practice. In seminar 1.3 on 18 November we test that insight against the practice of your own roles, strength and potential.
References
- Agrawal, A., Gans, J., & Goldfarb, A. (2018). Prediction machines: The simple economics of artificial intelligence. Boston, MA: Harvard Business Review Press.
- Agrawal, A., Gans, J., & Goldfarb, A. (2022). Power and prediction: The disruptive economics of artificial intelligence. Boston, MA: Harvard Business Review Press.
- Elish, M. C. (2019). Moral crumple zones: Cautionary tales in human-robot interaction. Engaging Science, Technology, and Society, 5, 40–60. https://doi.org/10.17351/ests2019.260
- Fayol, H. (1916). Administration industrielle et générale. Paris: Dunod.
- Haslam, S. A., Reicher, S. D., & Platow, M. J. (2011). The new psychology of leadership: Identity, influence and power. Hove: Psychology Press.
- Kets de Vries, M. F. R. (2001). The leadership mystique. London: Pearson.
- Lord, R. G., & Maher, K. J. (1991). Leadership and information processing: Linking perceptions and performance. Boston, MA: Unwin Hyman.
- Meindl, J. R., Ehrlich, S. B., & Dukerich, J. M. (1985). The romance of leadership. Administrative Science Quarterly, 30(1), 78–102.
- Mollick, E. (2024). Co-intelligence: Living and working with AI. New York, NY: Portfolio.
- Orlikowski, W. J. (2007). Sociomaterial practices: Exploring technology at work. Organization Studies, 28(9), 1435–1448. https://doi.org/10.1177/0170840607081138
- Popper, M., & Mayseless, O. (2003). Back to basics in leadership: The role of attachment in the leader–follower relationship. The Leadership Quarterly, 14(1), 41–65.
- Smircich, L., & Morgan, G. (1982). Leadership: The management of meaning. Journal of Applied Behavioral Science, 18(3), 257–273.
- Smith, B. C. (2019). The promise of artificial intelligence: Reckoning and judgment. Cambridge, MA: MIT Press.
- Stacey, R. D. (2001). Complex responsive processes in organizations. London: Routledge.
- Stogdill, R. M. (1948). Personal factors associated with leadership: A survey of the literature. Journal of Psychology, 25, 35–71.
- Uhl-Bien, M. (2006). Relational leadership theory: Exploring the social processes of leadership and organizing. The Leadership Quarterly, 17(6), 654–676.
- Uhl-Bien, M., & Marion, R. (2009). Complexity leadership theory. Organizational Dynamics, 38(2), 150–160.
- Weber, M. (1905). Die protestantische Ethik und der Geist des Kapitalismus. Archiv für Sozialwissenschaft und Sozialpolitik, 20–21.
- Weber, M. (1922). Wirtschaft und Gesellschaft. Tübingen: Mohr.
- Weick, K. E. (1995). Sensemaking in organizations. Thousand Oaks, CA: Sage.
- Zuboff, S. (1988). In the age of the smart machine: The future of work and power. New York, NY: Basic Books.