To Lead Is to Be Human

There is a particular anxiety moving through leadership in 2026, and it is worth naming plainly: a great deal of what managers were once valued for, a machine now does faster and, increasingly, better. It summarises the quarter, drafts the strategy memo, surfaces the risk in the numbers, suggests the decision. The work that filled a manager's week — the analysis, the synthesis, the reporting — is being quietly hollowed out. The honest question underneath the anxiety is not "how do I use these tools" but "what, exactly, am I now for?"

The field has an answer, and it is the right instinct pointed at the wrong target. The answer is human-centred leadership: as the technical work automates, the human work becomes the differentiator. What AI cannot do becomes what a leader is for. The instinct is sound. The trouble is in what "the human work" has been taken to mean. In most of its current form, human-centred leadership has been packaged as a set of behaviours — show empathy, listen actively, admit you don't have all the answers, check in on wellbeing — a competency bundle bolted onto the existing skills model and taught the same way. And a competency bundle is exactly the kind of thing that, in the end, a sufficiently capable machine can imitate. An assistant that opens every exchange with "that sounds really hard, how are you holding up?" is performing the behaviours of human-centred leadership. It is not being human, and people can tell.

This paper takes a harder line on what the human core actually is. It is not a set of warmer skills layered onto a leader. It is a way of being, and it is grounded in the one thing a machine does not have and a person cannot escape: the human condition. We are beings with something at stake. We are uncertain, exposed, mortal, capable of keeping or breaking our word, able to be genuinely present to one another or merely to transact. Leadership, stripped of the tasks now being automated, comes down to what a person does with that condition in front of others. That is not a soft addition to the job. As the rest of the job dissolves into the machine, it is becoming the job.

Picture the moment the abstraction becomes concrete. A leader stands in front of a team the morning after a third of their colleagues have been let go. There is no analysis to deliver and no framework that helps. What the room needs in that moment is not information — everyone has the information — and it is not optimism, which would insult them. What the room needs is a human being who will stand in front of them and not pretend: who is visibly affected and still steady, who says the true hard thing rather than the safe managed thing, who stays in the room with the discomfort rather than rushing through to next steps. A leader who can do that holds the team together. A leader who reads a word-perfect statement about resilience and synergy loses them — and loses them in proportion to how polished the performance is, because the polish is evidence of distance. Nothing in that moment is a skill in the trainable sense. It is entirely a matter of who the leader is able to be when there is nothing to hide behind. Hold onto that scene as the test case for everything that follows: not the easy moment where warmth is cheap, but the hard one where being genuinely human costs something and a performance of humanity makes it worse.

The dominant paradigm and why it hollows out

Human-centred leadership has, almost overnight, become the consensus frame for leading in the age of AI, and the consensus is converging on a recognisable list: empathy, authenticity, vulnerability, presence, emotional intelligence, psychological safety. The list is not wrong. Every item on it points at something real. The failure is in how the list is held — as a set of behaviours to perform, on the same logic that gave us the knowing–doing gap in the first place.

Three things go wrong when the human is treated as a skill bundle. The first is that it becomes performative. A behaviour learned as a technique, severed from the way of being that would make it genuine, reads as exactly that: a technique. The leader who has been trained to "show vulnerability" and dutifully shares a rehearsed weakness at the top of the offsite has not been vulnerable; they have performed a routine, and the room registers the performance and trusts them slightly less for it. Vulnerability done as a move is the opposite of vulnerability.

The second is that it becomes automatable. The whole argument for human-centred leadership is that it is the part machines cannot do. But if "the human part" is reduced to a script of empathetic phrases and check-in questions, then it is precisely the kind of pattern a machine can produce on demand — and increasingly does. By defining the human as a set of observable behaviours, the field has defined it as something a sufficiently good imitation can satisfy, which surrenders the very ground it was trying to defend.

Vulnerability done as a move is the opposite of vulnerability. People can feel the difference between a person being human and a person performing humanity.

The third is that it becomes exhausting and unsafe. Told to be more empathetic, more available, more vulnerable, without any change in the way they are being, leaders comply by simulating — manufacturing emotional labour they do not feel, over-disclosing because they have been told disclosure is the goal, absorbing everyone's distress as a performance of care. This is a fast route to depletion, and it has given human-centred leadership a deserved reputation in some quarters for producing burnt-out leaders and boundaryless cultures. The fault is not in the human core. It is in trying to reach it through behaviour rather than being.

The human core is a way of being

To get underneath the behaviour we have to ask what makes any of these qualities real when they are real. Why does one leader's admission of uncertainty draw a team closer while another's lands as weakness or manipulation? The words can be identical. The difference is not in the behaviour; it is in who the person is being as they do it — whether the act issues from a genuine relationship to their own exposure, or is reached for as a tactic. The human qualities that matter in leadership are not behaviours that can be specified and trained directly. They are expressions of a way of being, and they are only as real as that way of being is.

This is where the human condition becomes the subject rather than the backdrop. What gives a leader's presence its weight is that they, like everyone in the room, are a being with something genuinely at stake — with limited time, real uncertainty, the capacity to fail and to be seen failing, and the standing capacity to keep their word or break it. A machine has none of this. It risks nothing, loses nothing, and has no word of its own to honour. Its simulations of care are unfalsifiable performances with nothing behind them. The reason people reach for a human leader in the moments that matter — the loss, the crisis, the decision that cannot be reduced to data — is not that the human has better information. It is that the human has skin in the same game, and only a being with skin in the game can offer the things people actually need from a leader.

The model: the human core

The model below names three capacities that constitute the human core of leadership. Each is a way of being rather than a behaviour. Each produces a specific good that an organisation cannot manufacture any other way. And each is structurally unavailable to a machine, because each requires a being with something at stake.

Vulnerability — the capacity that produces trust

Vulnerability, in the research tradition that has done most to clarify it, is the willingness to show up and be seen when there is no guarantee of the outcome — to act under uncertainty, to take emotional risk, to be exposed. Brené Brown's body of work establishes it not as weakness but as the precondition for courage, connection, and trust: there is no daring without exposure to being wrong or unwanted. In a leader this is concrete. It is the willingness to say "I was wrong about that," to make a consequential decision without certainty and own it, to ask for help in front of people who report to you. None of this is a technique, and all of it generates the same thing: trust. People extend trust to a leader who is willing to be seen, because being seen is costly and the willingness to pay that cost is information about character that no amount of competent performance supplies. A machine cannot be vulnerable. It has nothing to lose by being wrong and no self to expose. It can output an apology; it cannot be sorry. The trust that vulnerability produces is therefore one of the human core's irreducible goods.

Integrity — the capacity that produces workability

Integrity here is meant not as a moral badge but in the sense the ontological tradition gives it: being whole and complete, honouring one's word as oneself, so that word and action are aligned. A leader with integrity is one whose commitments can be relied on to mean what they say — and the organisational consequence of that is workability. A team, a project, a system is workable to exactly the degree that the words spoken inside it can be trusted to correspond to what happens. Where integrity is thin, every commitment must be privately discounted, every plan hedged, every promise re-verified, and the cost of that discounting is the friction that grinds organisations down. A machine has no word of its own to keep; it has outputs, not commitments, and nothing is at stake for it in whether they hold. Integrity is a stand a being takes about their own word, and only a being can take it. The workability it produces is the second irreducible good.

Presence — the capacity that produces meaning

Presence is the capacity to actually be here, with another person, attention undivided — not the performance of attention but the fact of it. It is the rarest of the three and the most quietly decisive, because what people most need from a leader in the moments that count is the experience of being genuinely met. Presence is how a person comes to feel that their work, and they themselves, matter — which is to say, presence is how meaning gets made between people. A machine can simulate attention with uncanny fidelity; it can remember everything you said and reflect it back. What it cannot do is be present, because presence is not a behaviour to be reproduced but a being actually here, with stakes, choosing to attend to you rather than to the thousand other claims on a finite life. People feel the difference, and the difference is everything. The meaning that presence produces is the third irreducible good.

A machine can output an apology. It cannot be sorry. It can simulate attention. It cannot be present. The difference is the whole of the human core.

Run a single situation through the three. A leader's flagship initiative has just failed, publicly, and the organisation is watching how they handle it. At the level of vulnerability, the question is whether they can be seen in it — own the call that did not work, without spin, without the pre-emptive blaming of circumstance, without the carefully hedged non-apology — or whether the exposure is too much and they manage it into something defended and small. The team reads which it is instantly, and extends or withdraws trust accordingly. At the level of integrity, the question is whether the commitments the leader now makes about what changes can be believed — whether their word still means what it says after a failure, or has quietly become the inflated currency of someone protecting a position. The workability of everything that follows turns on it. And at the level of presence, the question is whether, in the conversations that come next with shaken people, the leader is actually there — attending to the person in front of them rather than to their own reputation management running in the background. People can feel which has their attention. A leader who meets the failure with genuine vulnerability, intact integrity, and real presence comes out of it more trusted than before; the failure becomes the occasion of their credibility rather than the end of it. A leader who performs all three comes out diminished, because a performance under exactly this much pressure is transparent. The situation does not test what they know. It tests who they are.

What changes specifically because of AI

It is worth being precise about why this matters now rather than as a perennial truth, because the human core has always mattered and the claim here is that something specific has shifted. Three dynamics are new. The first is saturation: the environment is now flooded with fluent simulations of the human — messages, summaries, even expressions of care — produced at scale by systems with nothing behind them. When competent, warm-sounding output becomes effectively free and infinite, its signalling value collapses, and the only thing left that signals anything is the genuine article, which is costly precisely because a real person had to actually mean it. Genuine humanity becomes the scarce signal in a sea of cheap imitation.

The second is the uncanny gap. As simulations improve, people become more rather than less attuned to the difference between being met by a person and being processed by a system, and they grow allergic to the performance of humanity exactly as it becomes more common. A leader who deploys the behaviours of human-centred leadership in this environment is now competing with machines that deploy those behaviours more smoothly — and losing, because the one thing the machine cannot add, actual presence behind the words, is the only thing anyone is really looking for. The behavioural version of human-centred leadership does not just fail to help here; it puts the leader in a competition with software that the software wins.

The third is the displacement of contact. As more interaction is mediated and automated, genuine human contact becomes scarcer in absolute terms, which raises its value to the people getting less of it. Put together, these three dynamics do not threaten the human core; they are a windfall for it. They strip the value out of everything the human core was competing against and concentrate it on the one offering a machine cannot counterfeit.

Why the value rises as automation advances

The ordinary fear is that automation diminishes the leader. The model implies the reverse. As machines take over the technical and informational work, two things happen at once. The relative scarcity of the human core increases — in a world of fluent automated competence, the things only a being can offer become the rare and distinguishing goods. And the demand for them rises, because the same automation that displaces tasks also unsettles people, thins out human contact, and floods the environment with plausible simulations, which makes genuine trust, real workability, and actual presence more valuable precisely as they become harder to find. The leader's worth does not survive automation by competing with it on the automatable. It rises by occupying the ground automation cannot reach. This is why the human core is not a nostalgic preference but a strategic position: it is the one source of leadership value that gets more valuable as the machines get better.

The human condition as the subject

It is worth dwelling on why the human condition, and not merely "people skills," is the right name for what is at issue, because the choice of name decides whether the whole thing is taken seriously. The qualities the model names are not arbitrary nice-to-haves; they are the specific human responses to the specific facts of being human, which is exactly why they cannot be detached from a being who actually faces those facts. Vulnerability is the human response to the fact that we are exposed — that we can be wrong, rejected, and seen failing, and cannot make ourselves immune to it. Integrity is the human response to the fact that we are the kind of being that gives its word, and that a self is constituted, over time, by the relationship it keeps to its own commitments. Presence is the human response to the fact that our time and attention are finite, so that to give them is to give something that cannot be got back.

Each capacity is a way of meeting a condition that a machine simply does not have: a machine is not exposed, gives no word that constitutes it, and spends no finite life in attending. This is why the human core cannot be transferred to a system however capable — and also why it cannot be faked for long by a person. A performance of vulnerability without real exposure, of integrity without a real relationship to one's word, of presence without the real expenditure of a finite life, is a performance with the load-bearing element removed, and under pressure it shows. The human condition is not the background to the model. It is the source of everything the model contains, and the reason its contents are scarce.

Why it cannot be proceduralised — and how it is developed

If the human core were a set of behaviours, it could be proceduralised, and then it could be automated, and the whole argument would collapse. It is not, and the proof is in what happens when organisations try. Mandated vulnerability produces performance. Scripted empathy produces the uncanny. Required presence produces the appearance of attention with none of its substance. Every attempt to reach the human core through behavioural specification destroys the thing it reaches for, because the thing is a way of being and behaviour is its expression, not its source.

This does not make the human core undevelopable. It makes it developable only in the way ways of being are developed: experientially, by working on who a leader is being rather than on what they do. Vulnerability is built not by instructing leaders to disclose but by working on their relationship to their own exposure — until being seen stops registering as a threat to manage and becomes available as a choice. Integrity is built not by exhorting people to keep their word but by making visible the cost of the commitments they have quietly broken, and restoring their relationship to their own word. Presence is built not by teaching attention as a technique but by working on what pulls a leader out of the room — the anxieties and distractions that make genuine attending feel unsafe or unaffordable. In each case the work is on the being, and the behaviour follows on its own, genuine because it is no longer manufactured.

This is demanding work, and it asks something of leaders that a content module never does: a willingness to be changed rather than merely informed. That is uncomfortable, and the discomfort is not incidental. The capacities at issue — exposure, the honouring of one's word, undivided attention — cannot be developed from a safe distance, because the distance is exactly what has to close. An organisation serious about the human core has to be willing to fund work that is experiential and personal rather than instructional, and to judge it by what changes in how its leaders are with people, not by what they can now say about being human.

Where organisations go wrong

The failure modes are predictable enough to list, and each is a version of the same mistake: trying to get the good the human core produces without doing the work on being that produces it. The first is the mandated-vulnerability programme — the off-site where leaders are required, in sequence, to share a personal struggle, producing a room full of rehearsed disclosures that perform openness and deepen no trust, because nothing in the leaders' actual relationship to their own exposure has changed. The second is wellness theatre — the apparatus of check-ins, resources and empathy training that lets an organisation feel it has addressed the human while leaving untouched the actual ways of being that determine whether people are trusted and their work made to matter. The third is the competency-isation of the human: writing vulnerability, authenticity and presence into the leadership framework as rated behaviours, assessing leaders against them, and thereby guaranteeing that leaders perform the rated behaviours rather than develop the ways of being — optimising, precisely, for the simulation.

Each of these is well-intentioned and each makes the underlying problem worse, because each teaches leaders that the human is a set of moves to be executed, which is the belief that produces the hollow, performed humanity people have already learned to distrust. The organisations that get this right do something harder and less legible: they invest in genuine development of the observer their leaders are, and they tolerate the discomfort and the slower, less surveyable progress that real change of being requires. It is a worse fit for a procurement process and a better fit for reality.

Making the human core observable

Because the human core is a way of being, it cannot be measured by a knowledge test — but the goods it produces are observable, and that is where assessment belongs. Trust shows up in measurable behaviour: whether people bring problems early or hide them until they explode, whether bad news travels up or gets trapped, whether discretionary effort is offered or withheld. Workability shows up in how reliably commitments inside a team correspond to outcomes — the gap between what is promised and what happens, the volume of re-verification a system carries because its words cannot be trusted. Meaning shows up in whether people experience their work as mattering, which surfaces in retention, in the quality of attention people bring, in whether they stay through a hard patch or leave at the first better offer. None of these measure vulnerability, integrity, or presence directly; they measure the goods those capacities produce, which is the right place to look. When the human core is genuine, these markers move together. When it is performed, they do not move at all — the leader runs the empathetic script and the trust, the workability, the meaning stay exactly where they were.

What this means for developing leaders

If the human core is a way of being rather than a set of behaviours, the implications for how organisations develop leaders are large and largely unwelcome, because they cut against how the development function is built. The first is a reallocation. The marginal hour and the marginal pound currently go, overwhelmingly, to content — models of empathy, frameworks for difficult conversations, the curriculum of human-centred leadership delivered as knowledge. On the argument of this paper, that spend is buying the wrong thing: it enriches what leaders know about being human while leaving untouched whether they are being human, which is the only variable that produces the goods. The leverage is in experiential work on the observer, which is exactly where conventional programmes spend least, partly because it is harder to design and partly because it is harder to defend in a budget meeting.

The second implication is a change in the question an organisation asks of any leadership intervention. The usual question — what will our leaders learn? — is a layer-one question, and an intervention that can only answer it will produce more articulate, no more human leaders. The better question is: what will our leaders be able to be, under pressure, that they cannot be now, and how will we see it in the trust, the workability, and the meaning around them? An intervention that cannot answer the second question is selling content about the human, not development of it, however warm its brochure.

The third implication is patience, and a tolerance for the illegible. Change in a way of being does not happen on a content schedule and does not show up cleanly on a satisfaction form the following week. It shows up later, in how a leader is with people when something goes wrong, in whether bad news reaches them, in whether their team's discretionary effort rises. An organisation serious about the human core has to be willing to fund work whose results are real but slow and indirect, and to resist the pull back toward the legible, surveyable, and ultimately hollow.

For public institutions and governments the stakes are sharper still, because the human core is doing work there that nothing else can. Public trust, the willingness of people to be governed and served by an institution, rests on the felt sense that real human beings with integrity stand behind it — and in an age of automated, mediated, and frequently distrusted public communication, leaders who can be genuinely present and genuinely accountable are not a soft adornment to public service. They are increasingly the thing that holds its legitimacy together. The argument that the human core is a strategic asset is, in the public sphere, also an argument that it is a civic one.

Two objections

The first: isn't this just emotional intelligence in new clothing? Emotional intelligence, as usually taught, is a set of competencies — recognise emotions, regulate them, read the room — and as such it sits squarely in the behavioural paradigm this paper is arguing against. The human core is not a more refined competency set; it is the way of being that determines whether any such competency is genuine or performed. A leader can score highly on an emotional-intelligence assessment and still be running a sophisticated performance of humanity with no one actually home. The distinction is precisely the one between knowing about and being — and it is the distinction emotional-intelligence training, conducted as content, tends to miss.

The second: doesn't this romanticise the human and risk pushing leaders into over-exposure and burnout? It would, if the human core were a demand for maximum disclosure and limitless availability — but that is the behavioural misreading again. Vulnerability is the willingness to be seen when it serves, not compulsive self-revelation; presence is undivided attention when one is with someone, not perpetual availability to everyone. Grounded in a genuine way of being rather than a performance of care, the human core is in fact protective against burnout, because the depletion comes from manufacturing feeling one does not have. The leader who is actually present for an hour is less depleted than the one performing availability for twelve. The risk the objection names is real, and it is a risk of the very behavioural approach this paper rejects.

Conclusion

The age of automation is not making leaders less necessary. It is stripping away everything about the role that was never the point, and leaving exposed the part that always was: a person, with something at stake, being human in front of others in a way that produces trust, makes the organisation workable, and lets the work mean something. That part cannot be automated, because it is not a behaviour to reproduce but a way of being — and a machine, having nothing to lose and no word to keep and no finite life from which to give its attention, cannot be in that way at all. To lead, when the machines have taken the rest, is to be human on purpose and in public. It is the oldest thing leadership ever asked, and it is about to become the only thing it asks that matters. There is an irony worth sitting with in that: the most advanced technology ever built is, by taking over everything a leader could be replaced at, returning leadership to its least technological core — to the wholly human business of being trusted, being reliable in one's word, and being genuinely present to other people. No tool delivers those, and no tool ever will, because they are not deliverables; they are the expression of a being who has something at stake and chooses how to meet it in front of others. The leaders who understand this will stop competing with the machine on the ground it owns and invest in the ground it cannot touch — which happens also to be the ground on which leadership was always, in the end, decided. Developing leaders equal to it — and being able to show that the trust, the workability, and the meaning actually moved — is the work this desk exists to advance.