The model is not the leader

“The model recommended it.” It is becoming one of the most convenient sentences in modern management. It sounds like an explanation. More often it marks the place where an explanation should have been.

That is not an argument against the tools. Used well, AI widens what a leadership team can see before it decides. It reads faster than any of us, connects material we would never have placed side by side, and drafts the counter-argument we were too invested to write. Declining that out of nostalgia only makes an organisation slower and less informed.

The temptation lies elsewhere: to mistake a better answer for a better decision — intelligence for judgement. Intelligence, here, is the capacity to produce a plausible, well-reasoned response to a question. Judgement is the capacity to know which question the situation is really asking, what the answer leaves out, and what it will cost the people who live inside it.

I was reminded of the difference in an assessment centre for a C-level appointment. Three candidates presented their recommendations on the same assignment, and the three presentations were close to identical — in structure, in argument and in conclusion. The difficulty began with the questions. Why this route rather than another? What would follow from it in the short term, the medium term and the long term? None of the three could defend the reasoning behind their own recommendation. None of them progressed.

I do not know how those presentations were prepared, and I will not guess. What the room showed was that an answer can be delivered fluently by someone who has not yet made it their own — and that at this level the answer is never what is being assessed. The capacity to stand behind it is.

Leadership decisions face the same test. Consider a restructuring. A model can show that the new structure removes duplication and that the savings are real. It cannot sit across the table from the person whose role disappears, or know that the people watching how he is treated will decide from it whether to stay. That knowledge lives in the room, not in the data.

There is early evidence that the distinction erodes quietly. In a 2025 survey of 319 knowledge workers, researchers at Microsoft Research and Carnegie Mellon University found that higher confidence in generative AI was associated with less self-reported critical thinking, while higher confidence in one’s own ability was associated with more. The study measures what people report, not what they do, and it does not show that AI causes the change. But it describes a pattern most leaders will recognise: the more we trust the answer, the less we examine it.

A separate 2025 experiment points the same way. Participants given faulty AI support answered fewer than half as many reasoning questions correctly as those working without it. A warning on screen helped, but did not close the gap.

CURRENT describes leadership through three movements — Stay, Sense and Shape — and each is a place where the model’s answer has to meet a person. To Stay is to remain with the question long enough not to mistake the first fluent answer for the end of thinking. To Sense is to read what the answer cannot carry: the context, the people, what was left out. To Shape is to act and to own what follows, because the organisation will respond to the decision, not to the recommendation.

A recommendation can inform a decision. It cannot carry it. Accountability belongs to a person who can be asked why, and who lives with what followed. The NIST AI Risk Management Framework treats governance — who is responsible for what — as a core function, not a footnote to the technology.

None of this argues for less AI. It argues for a different relationship with it. The model is at its most valuable as a thinking partner when it is contested — asked what it assumed, what it left out, and what would have to be true for it to be wrong.

The question, then, is not whether the model is intelligent. It is whether we will still recognise the difference between a good answer and a decision we are prepared to own.