Judgment Was the Boss’s Job.
AI Made It Everyone’s.

Picture the last hard call you made. Not a routine one. A real one, where money or direction or someone's job was riding on it. Now picture the people who were in the room when you made it. Here is the part worth sitting with. The person who understood the thing you were deciding about best, the one closest to the actual mechanics of it, was almost certainly not you. They were three levels down. They might not have been in the room at all.

For most of the history of running organizations, that was not the case. The person at the top of the table was supposed to be the person who knew the most. That is the model nearly all of us were trained on, whether anyone said it out loud or not. You set the direction. You hold the plan. You make the call. And the organization moves on your confidence, because your confidence was, for a long time, an honest signal that you knew more than anyone around you.

That signal has stopped being honest. And almost nobody has told you, because nothing inside the old model raises its hand when it breaks. That is what I want to walk through today. Not whether to use AI, or how fast. The quieter question underneath all of that. What does your actual job become, once your confidence is no longer the thing that makes the organization work?

Let me give you the number that made this concrete for me, because it is not a small drift, it is most of the room. This spring, LinkedIn surveyed more than twelve hundred C-suite leaders. Half of them said they have no clear view of the roles and skills their organizations are going to need as AI matures. And seventy-eight percent said they are moving on AI faster than they can measure. So if you have felt that, the sense of moving quickly on something you cannot fully see the shape of, you should know that is not a personal failing and it is not you falling behind. That is the normal condition of senior leadership right now. Roughly four out of five of your peers are in exactly the same position. The ground moved, and the instruments most of us were handed do not read this terrain.

So let me back all the way up, because to see why the ground moved, you have to see what the old ground was made of.

Think about where our whole idea of leadership comes from. The pictures are old and they are vivid. The ship's captain who had sailed seas his crew had never seen. The general standing over a map of a battlefield no individual soldier could ever take in. The founder who held the whole shape of the enterprise in his head while each worker stood at one bench, doing one thing, seeing one part. Those figures gave us the first real theory of leadership. Scholars call it the Great Man theory, and the idea is right there in the name. Leadership is a property of one exceptional individual. The organization rises or falls on the person who can see furthest.

And here is what is interesting. For about a century afterward, the field kept refining that idea without ever actually leaving it. First came trait theory, which asked what qualities the great leader is made of. Then behavioral theory, which said it is less about who the leader is and more about what the leader does. Then situational leadership, which is probably the most sophisticated version, and the one a lot of executive training still runs on. Situational leadership says the right thing to do depends on the conditions. The good leader reads the situation in front of them and adapts their behavior to it.

Notice what every single one of those has in common. They all keep the individual leader at the dead center of the picture. They argue about what the leader should be like, or what the leader should do, or how the leader should flex. Not one of them ever stops to ask a stranger question, which is whether leadership has to live inside one person at all.

And the reason none of them asked is that they did not need to. The whole tradition was resting on something that used to be solidly true. The leader genuinely was the apex of the organization's information. They really did know more. The captain really had sailed the sea. The confidence was earned because the knowledge was actually there, sitting behind it.

Here is where it turns. That assumption, the one holding up every version of the model, the leader sits at the center with the longest view, has quietly flipped upside down.

In a modern knowledge organization, the people below you hold more of the information that actually bears on your decision than you do. Not all of the information. More of the decision-relevant information. The specialist three levels down understands the system your call is about to touch in a way you could not reconstruct in a week. The team sitting closest to the customer can see what your strategy is actually hitting when it lands, while you are still looking at the deck that described it. The knowledge did not disappear. It redistributed. It spread out and down and sideways into the organization, and it mostly stopped living at the top.

And the model kept right on running, because here is the cruel mechanical detail. Projecting certainty feels exactly the same from the inside whether or not the knowledge is still behind it. The experience of being confident does not change when the thing justifying your confidence quietly leaves. There is no internal alarm. No light comes on. You stand up in the meeting, you make the call with the same steady voice you have always used, and it feels like leadership, because it has the same texture as the leadership that used to work. The form survived. What was underneath it left without a sound.

Now, this had been happening slowly for years. AI is what made it impossible to keep ignoring, and it did that in a very specific way. AI dropped the cost of producing finished, confident-looking work down to almost nothing.

Sit with what that does. Harvard Business Review put it well this June. They said the first draft is now the easy part. The hard part, the genuinely hard part, is judging whether the polished thing in front of you is any good. And they made a point that I think a lot of organizations are about to learn the expensive way. Training everyone to be fluent with the tools is necessary, and it is nowhere near enough. The fluency is the cheap part. The judgment is the scarce part.

Think about what that means for the oldest signal we have. For a hundred years, if someone handed you a sharp, polished, well-argued strategy memo, the polish told you something. It told you a person with real competence spent real time and real thought producing it. The artifact was evidence of the judgment behind it. Now anyone can generate a fluent strategy memo, a clean board deck, a tight competitive analysis, in about a minute. So the polish has gone silent. It no longer tells you anything about the thinking behind it, because there may not have been very much thinking behind it at all. The thing that used to be a proxy for judgment has been completely severed from judgment.

Now here is the part most people miss, and it is the whole reason I wanted to talk through this rather than just write it down.

Step back from yourself for a second. Stop thinking about the leader. Because the exact same shift is running through every single level of your organization at the same time, and that is the thing that actually changes what you do.

The industrial model we all inherited divided the world cleanly. Judgment lived at the top. The executives and the board, they did the deciding. They figured out what was worth doing. Everyone below that line was hired, measured, paid, and promoted on execution. On doing the thing well and fast. And that division was not arbitrary. It made complete sense, for one reason. Execution was the scarce, expensive, valuable thing. It cost a fortune to make things and move things and produce things, so naturally you organized the whole company around producing efficiently, and you kept the small expensive supply of judgment locked up where the big decisions got made.

AI dissolves exactly that arrangement. Because as execution gets cheap, and it is getting cheap fast, at every level, the thing that is still scarce, the thing that still has to be supplied by a human at every level, is judgment. The cognition that decides what is actually worth doing in the first place, and then looks at the output and knows whether it is any good or whether it just looks good.

So the capacity that the organization only ever built for the corner office, the deciding, the discernment, the knowing-what-matters, now has to become the working skill of the entire place. Your analyst needs it. Your engineer needs it. The person who just generated four polished options in an afternoon needs to be the person who can tell which one is right, and that is a judgment skill, and almost no one below the executive line was ever hired or trained or measured for it.

And here is the move that everything else depends on. The instinct in every organization right now is to treat this as a technology problem. Which tools, which budget, which rollout, which training. I want to suggest it is not a technology problem at its root. It is a problem of cognition. It is about where thinking happens in your organization, and who is allowed to do it, and whether the people who now have to exercise judgment were ever set up to.

I will be honest about why I see it that way, because it is relevant. I am autistic. None of the cognition that most people get for free ever came automatically to me. I have had to design how my own thinking works, deliberately, out in the open, piece by piece, because the automatic version was never available. And when you have had to build your own cognition by hand, you stop seeing thinking as a thing that just happens inside a person, and you start seeing it as something with a structure, something that can be well-designed or badly-designed, something you can build conditions around. That is the lens I am bringing to your organization. The reason I can tell you this is a cognition problem and not a technology problem is that cognitive architecture is the thing I have spent my life having to be explicit about. So when the whole working world suddenly has to make judgment a designed capability instead of an inherited one, that is, to me, the obvious frame. It is the only frame I have ever had.

There is research catching up to this, by the way. HRZone made the point this spring that an AI transformation is not only a systems event, it is a brain event, and organizations are missing the returns because they keep overlooking the human cognitive infrastructure the whole change is supposed to run through. And LinkedIn's own chief business officer said something even sharper. The top-down transformation mandate, the executive standing up and ordering the change from the top, may actually be making it worse. Because this change cannot be ordered down a hierarchy. The judgment cannot be installed by decree. It has to be built into how the place thinks.

So let me put the whole shift into one sentence, and then I will tell you what to do about it.

A leader's confidence used to be an accurate readout of knowing more than everyone in the room. That readout has broken. Your job is no longer to be the best-informed person in your organization. Your job is to design how your organization comes to know.

Alright. If you are no longer the apex, if the job that used to sit on top of all that information has genuinely changed shape, then the obvious question is the practical one. What replaces it? What do you actually do on Monday?

I think the new job is three moves. And I want to walk you through each one with a real situation, because in the abstract they sound like management slides, and they are not.

The first move is to gather the signal your organization already holds.

Here is the situation. There is an engineer, three levels down from where the roadmap gets decided. And she spots something. She sees an integration risk, a real one, the kind of thing that turns into a six-month problem if it ships wrong. So she does the right thing. She flags it. She writes it up clearly in a thread. And then that thread, and that warning, just never travels. It gets buried under forty other messages. It never reaches the people who are about to commit the roadmap. And so the decision gets made cleanly, confidently, in a room where the single most important piece of information was sitting in a thread nobody upstream ever read. The signal existed. It died in transit.

Your old job, in that situation, was to be the one with the answer. To know about the integration risk yourself, somehow, and to raise it. Your new job is different and frankly more useful. Your job is to build the path. To make sure that when an engineer three levels down sees something real, there is a route that carries that signal up into the room intact, before anyone commits, every single time, without depending on her happening to catch the right person in the hallway. You are not supplying the knowledge anymore. You are building the channel the knowledge travels through.

The second move is to place the judgment.

Different situation. There is a pricing decision on the table. And it lands on the desk of the senior vice president, because she is the most senior person attached to it. That is just how it works. Except the actual demand modeling, the deep work that the decision should turn on, was done by an analyst two levels below her. And nobody, anywhere in this process, was actually named as the person who owns the result of this call. So watch what happens. Three different things, holding the relevant knowledge, making the actual decision, and owning whether it works out, all collapse onto one person by sheer organizational default. The SVP holds the title, so she gets all three, even though the knowledge lives somewhere else and the ownership was never assigned to anyone on purpose.

MIT studied this. They looked at enterprise AI efforts, the ones that actually paid off versus the ones that quietly failed. And the cleanest predictor was almost boringly simple. The ones that worked had one clear, accountable human owner. A name. The ones that failed were owned by a committee, or owned by the tool itself, or owned by no one at all.

So placing the judgment means doing on purpose what otherwise happens by accident. It means deciding who holds the relevant knowledge, who actually makes the call, and who owns the outcome, and being willing to make those three different people. And it includes, now, deciding the machine's role explicitly. Where does the AI inform the decision. Where does it get to recommend. And where is it kept entirely out of the room. You assign those roles deliberately, instead of letting the org chart assign them for you while you are not looking.

The third move is to route the work so judgment keeps up with the pace.

Last situation, and you have probably already lived this one. Your team hands you a board memo. It is clean. It is sharp. It reads like something that took a careful week. And it took twenty minutes, because most of it was generated. Now, somewhere in those twenty minutes, a decision got made, or much more likely did not get made, about which paragraphs in that memo actually needed a human being to think hard, and which ones the machine could just finish on its own. If nobody made that call, then the memo is uniform on the surface and wildly uneven underneath. Some of it is load-bearing human judgment and some of it is fluent autocomplete, and they look identical.

Routing the work means making that call on purpose, task by task. For this piece of work, what stays fully human, because the judgment in it is the whole point. What does the machine amplify, where a person leads and the machine extends their reach. And what does the machine just absorb entirely, the genuinely rote part, freeing the human up for the parts that need them. That routing decision, repeated across everything your team produces, is now a core part of the job. Pace without routing just means you ship your blind spots faster.

Now step back and look at those three moves together, because here is what they actually are.

Remember the whole history we walked through. Great Man, trait, behavioral, situational. Every one of those theories did the same thing. It located leadership inside the individual leader. It just kept arguing about what that individual should be like. These three moves do something the entire tradition never did. They take leadership out of the person and relocate it into the organization's whole cognitive capacity.

Leadership becomes designing judgment as a capability that lives at every level, instead of reserving it for the top and assigning everyone else to execution. And you can hear how that completes the line the tradition was already walking. Situational leadership told you to read the conditions and adjust your own behavior to them. This tells you to build the conditions, so that good judgment lives wherever your organization actually needs it, whether or not you are standing there.

Engineers have a name for this, and I find it clarifying. They call it emergent design. The idea is that for a sufficiently complex system, you do not hand-build the final output yourself. You cannot. It is too big and it is moving too fast. Instead you set the conditions and the rules that all the parts run under, and the result emerges from those conditions. For a leader, that is not abstract at all. The conditions you set are concrete. The structure of the meeting. Who actually holds the decision rights. The rule for when the machine gets a vote and when it does not. The path that signal has to travel before any call gets made. You are not deciding every answer anymore. You are designing the system that produces good answers without you in the middle of each one.

And if you have been following this series, here is something you might find useful. Those three moves are not just a list I made up. They are the next three things I am going to build out, one at a time.

Gathering the signal is a piece of work I call the Cognitive Translation Protocol. It is about where signal gets lost between people who genuinely process information differently, and how you keep it intact in transit. Placing the judgment is DecisionOS. It is about exactly that failure where holding the knowledge, making the call, and owning the result get bundled onto one person, and how you pull them apart on purpose. And routing the work is the AI Cognitive Strategy Matrix. It is the task-by-task discipline of deciding what stays human, what gets amplified, and what gets absorbed.

And the order matters, so let me say why. A decision that gets made on signal that never arrived clean, and then gets scaled up by a system that produces faster than anyone can possibly check it, is the exact failure you already recognize from your own week. So you fix it in that order. Get the signal in first. Then decide who owns the call. Then, only then, decide how fast you let the work run.

I want to leave you with something practical, in a form I am going to come back to in every one of these. I think of it as an operating brief, and it has four parts. Let me just talk you through it.

It starts with the moment. The moment is this. A real decision is sitting in front of you, and the people who understand it best are not you.

Then there is the default, the thing that happens if you just lead the way you were trained to. The signal that would have corrected you never reaches the room. And the gap, the place where you were quietly wrong, ships out into the world at the full speed of your conviction. That is the actual cost of leading from your own read and letting the fluency of the output stand in for whether it was right. It is not that you freeze up. It is that you move fast and confident in a direction nobody got the chance to correct.

Then the redesign, the thing you do instead. You make your first job, before you supply a single answer, gathering the signal your organization already holds. You go get the read from the room before you give yours.

And then the test, the thing you can actually do this week. In your very next decision meeting, just watch for one moment. Watch for the moment the room shifts from figuring out what is true to selling what you have already concluded. It happens in almost every meeting, and once you see it you cannot unsee it. And then notice, honestly, how much of the hour fell on each side of that line. How much was discovery, and how much was just you, and everyone reading you, building the case for where you had already landed.

I want to be straight with you about what this is and is not. This is not a transformation program. It does not need a budget. It does not need a reorg. It does not need you to stand up at an all-hands and announce a new way of working. It is the first step onto the side of the job where you design the architecture deliberately, instead of just inheriting the one you were handed.

And I will tell you the strange part, the part nobody warns you about. For a while, doing this well is going to feel like doing less. Because you are deciding fewer of the answers yourself. You are sitting in meetings letting the read come up from the room instead of leading with yours. It can feel like you have gone quiet. Like you are contributing less.

You are not. You are doing the part of the job that does not leave when the execution leaves. The deciding-what-is-worth-doing. The knowing-whether-it-is-any-good. The building of a system that knows. Gather the signal. Place the judgment. Route the work, so that thinking keeps up with the speed.

So here is the one thing I would ask you to actually do. Just the test. In your next real decision meeting, find that moment where the room stops discovering and starts selling, and notice which side most of the hour lived on. That is it. That is the whole assignment. You do not have to fix anything yet. You just have to see it.

And if it is useful, stay with the series, because the next three pieces take these three moves and build each one out properly, one at a time. Signal first. Then ownership. Then pace.

Thanks for spending this time with me. I will see you in the next one.