If you have watched AI absorb your team’s tactical work and cannot precisely articulate what your job is now supposed to be, this piece is for you. The role is not disappearing. It is shifting to a specific territory that most executives have never been developed for, and the pipeline for the capability that matters most is narrowing at exactly the wrong moment.
I spent twelve months as a tank commander.
Long enough to understand the decision architecture from the inside. Long enough to know that the doctrine isn’t a constraint on judgment. It is the structure that makes judgment possible under pressure. A tank crew operates inside a command framework that has been stress-tested across a century of mechanized warfare and refined to the point of being almost algorithmic in its clarity. Who decides what, under what conditions, with what information, in what time window. The doctrine doesn’t leave much to interpretation, because interpretation under fire costs lives.
What struck me, years later when AI began arriving in enterprise organizations, was the structural similarity. Both systems involve a human at the top of a decision hierarchy whose primary value is not execution (the execution is handled by systems optimized for it) but judgment. Specifically, the judgment calls that the optimized systems cannot make. The ethical boundary, the contextual exception, the call that requires understanding what the numbers don’t contain.
The tank commander’s most important skill is not gunnery. It is knowing when not to fire.
As AI executes an increasing share of tactical and analytical outputs, the executive’s value shifts entirely to that same territory. The 20% of decisions that AI cannot make well. Not because it lacks data, but because they require the kind of judgment that emerges from accountability, relationship, and consequence. The question is whether most executives know what that 20% actually is.
“The tank commander’s most important skill is not gunnery. It is knowing when not to fire.”
The tactical layer is gone
This is not a future scenario. For many knowledge work organizations it is the current state.
The tactical layer (research synthesis, draft generation, pattern recognition across large data sets, scheduling optimization, performance analysis, code generation) is being absorbed by AI at a pace that is compressing job roles faster than most organizations have updated their understanding of what those roles are for.
The junior analyst who spent three years learning to synthesize market research is now producing that synthesis in hours. The question nobody is asking loudly enough is: what were those three years actually for? If the output was the point, the three years are no longer necessary. If the output was the training ground for something else (judgment, pattern recognition, the ability to assess whether a synthesis is correct) then removing the three years removes the development path for the capability that the senior role requires.
Most organizations are compressing that development path right now, in real time, without having decided whether they can afford to.
What remains
When the tactical layer is largely automated, what remains in the human executive role is precisely the territory that has always been hardest to define and easiest to undervalue.
Ethical judgment under genuine ambiguity. Not the cases where the ethical answer is clear and the AI can be instructed to optimize for it, but the cases where the values genuinely conflict and someone has to choose which one yields. These cases are rare. They are also the ones that define an organization’s character and, in the long run, its survival.
Relationship accountability. The commitment that exists between people who share consequences, not just information. AI can model relationships. It cannot be in one. The trust that a client places in a leader who has been through a crisis with them, the team’s confidence in a judgment call from someone who has demonstrated over time that they absorb the cost of being wrong. These are not products of data processing.
Psychological presence under pressure. The capacity to remain clear, steady, and honest when the environment is demanding the opposite. This is the tank commander skill. It is also the skill that the Architecture of Laziness piece is about, and the Vulnerability Narrative, and the Oxygen Thief. It is the meta-capability that makes all the other framework elements possible. And it is the one that no AI will develop, because it requires being affected by consequences in a way that only having them matters.
“AI can model relationships. It cannot be in one.”
The leadership curriculum that doesn’t exist
The problem is not that executives don’t understand AI. Most senior leaders at this point have a reasonable operational understanding of what AI can and cannot do. The problem is that almost no organization has asked the next-order question. If AI is doing more of the work, what are we doing to develop the specifically human capabilities that remain?
The answer, in most cases, is nothing systematic. The development path for ethical judgment, relationship accountability, and psychological presence is assumed to happen through experience. The same experience that AI is now shortcutting for junior talent. The pipeline for the capabilities that matter most in the AI-integrated organization is narrowing at exactly the moment those capabilities are becoming scarcer and more valuable.
This is the leadership curriculum that doesn’t exist. Not AI literacy. Every corporate learning team has an AI literacy program. The curriculum for becoming the kind of person whose judgment is worth something precisely because it cannot be replicated, automated, or optimized. The curriculum for the territory that is uniquely human not because humans are better at it, but because being human is what makes the judgment binding.
The tank commander doesn’t need to be a better gunner than the targeting system. They need to be the person who decides when the targeting system should not fire. That decision cannot be delegated. It cannot be optimized. It can only be made by someone with the judgment, the accountability, and the presence to bear its weight.
That is what the executive role is becoming. The question is whether the people filling it know that yet.
THE TEAM TEST
Name the three decisions your organization made in the last quarter that required specifically human judgment. Not better data, not faster processing, but the kind of judgment that comes from accountability and consequence.
Now ask: who in your organization is being developed to make those decisions? Not trained on AI tools. Developed as a human being, in the domains of ethical reasoning, relational accountability, and psychological presence under pressure.
If the answer is unclear, the bottleneck in your AI-integrated organization is not the AI. It is the humans who are supposed to be doing what the AI cannot.
That is the gap that no algorithm closes. It is also the only gap that matters.
