AI Is Not the Challenge. Leadership Is.

Some pictures stay with you. Lately, mine is a room I have never been in.

28 February 2026.

The first morning of Operation Epic Fury. In twenty-four hours, the United States struck more than a thousand targets in Iran.* Work that once needed stacks of paper, long meetings and many days was done in a single night. According to later reports, Palantir's Maven Smart System sat behind that speed, and inside Maven sat Anthropic's Claude, reading satellite images, signals and old files, then suggesting targets and putting them in order faster than any room full of analysts could.

Imagine that room. Dark screens. Lists that refresh on their own. Finding a target, ranking it and striking it used to take weeks. That night, it took hours.

I did not want to use a war to explain a problem at work. War is measured in lives, and work in time and money. But this is the one event whose consequences almost all of us are living with, and it is the clearest example I know of a tool moving faster than the people using it.

It is a marvel, and I will not pretend otherwise. I was a teenager when the Gulf War played out on our television screen, and many of the wars I have watched since then went on for years. When AI arrived in this one, many of us quietly assumed that speed would bring a quick ending. Six months later, the conflict continues, and the decisions have become harder, not easier.

The hard questions never appeared on any screen. Should we go to war at all? What should the morning after look like? Who is still in the room when speed and wisdom point in different directions?

These questions sit with leaders, and they always have. The easy answer is to say that the machine is the danger. I beg to differ. We keep staring at the knife, when the story was always about the hand that holds it.

A tool makes a job faster, sharper or cheaper, but only a job we already know how to do. A leader decides which job is worth doing, and then lives with that decision for years. Efficiency comes from the past. Value comes from someone taking responsibility for a future that has not happened yet. When we mix up the two, the organisation becomes faster but rarely becomes better.

AI is a tool. It sharpens the knife, and in some kitchens it even makes the cut. That is useful. But it is not rare, and any cook will tell you that a sharp knife cuts in both directions.

Three things follow from this.

The first is that AI is excellent at yesterday. Many organisations waste years forgetting their own history, so a system that hands that history back in seconds is a real gift. But memory is a poor guide to the future. When a leader treats a model trained on last year as a plan for next year, it is like driving a car forwards while looking only at the rear-view mirror. The road behind is very clear. The road ahead is still a decision.
The second comes from my own research. While writing my book, I spoke to a dozen leaders across industries, company sizes and continents, and when I asked how AI had helped, every one of them told me the same thing: it had lowered cost. Not one of them could show me a better decision that had come solely due to AI. Not one could show me new revenue that had come directly from it. Saving money lifts this year's profit, and that matters. But it gives no customer a reason to choose you, or to pay you more.
The third is that everyone now has the same knife. You can already see the result on LinkedIn, where many posts read like one relative forwarding the same good-morning message to every family group: the same rhythm, the same borrowed warmth, the same shine. That is what a common knife does to a voice. Now imagine two rival companies asking the same model the same question on the same Monday morning. By Friday, both have a strategy, and the two strategies look almost identical.

The leader's real job is the part that no machine can provide and no competitor can copy: the judgement about which future is worth betting on. When everyone can buy the same knife, the value cannot be in the steel.

Now walk into a restaurant kitchen.

A new knife arrives. It stays sharp, and an onion falls apart as if it had been waiting for this blade. By the next evening, every cook wants it, for bread, bone, fish and fruit alike. Plates leave the kitchen faster than ever, and the owner loves the numbers.

Out in the dining room, nothing has changed. Nobody orders the dish a second time. Nobody offers to pay more. A leader who watches only the orders leaving the kitchen, and never the faces at the tables, will miss this completely.

Then, at the new speed, a cut goes slightly wrong. A piece of fish cut in the wrong place, a plate that should have been stopped. Three tables eat the mistake before anyone notices. The knife does not care, and it will be just as eager on the next order. After a while, the mistake is forgotten. What remains is a more uncomfortable question: whose hand made that cut? The knife suggested. The cook followed. The person at the counter assumed someone else had already checked.

The search always ends at the hand. The new knife can sit in the drawer with the old ones. It cannot replace the chef who decides what goes on the menu, and whether the guest will ever come back.

That kitchen and the room on 28 February are the same picture. One is the size of a restaurant, and the other is the size of a country. A system can walk a team through its options at machine speed. It cannot tell leaders what they are willing to become in order to use that speed. A sharper tool never takes the decision away from a leader. It only makes an empty chair easier to see.

So when I say AI is not working at work, this is what I mean. The tool is working. The leadership around it has not caught up.

I have no quarrel with fluency. My quarrel is with fluency being mistaken for direction. An organisation that treats the model as its strategy becomes a faster copy of itself, speaking in a voice that could belong to anyone.

The first night in February did not make the chef unnecessary. It showed how quickly everyone notices that the chef is missing, once the knife can move on its own.

Question for you : What is one decision AI sped up in your organisation this year that nobody clearly owned?

Why AI Isn't Working at Work is available on Amazon - https://linktwin.co/ygozka

Get more information: https://abcatwork.org/

Source :The Washington Post, "Pentagon leverages AI in Iran strikes amid feud with Anthropic" (4 March 2026). This is the report on Claude working inside Maven and on more than 1,000 targets in the first 24 hours.