"We are hiring a Chief AI Officer. Don't look for a top AI expert."
Michael Maeder, a partner in executive search at August Leadership, reports this as the clearest instruction he received from a client, and he is right to have treated it as the clue it was. Most search assignments are order-filling. A specification arrives, the market is combed for people who resemble it, and the resemblance is mistaken for suitability—a process that has been described, accurately, as a machine for converting institutional anxiety into a shortlist. That this brief was interrogated rather than executed makes it unusual for an industry that over indexes on speed to fill, and the interrogation produced something genuinely useful. The client's pilots had worked. What frightened them was not failure but the far commoner fate of the successful experiment that never becomes operating reality. The technical capability was already in the building. What was absent was the capacity to prioritise across competing claims, to enforce a stop decision, and to hold anyone to a horizon beyond the next quarter.
That is a good diagnosis. It is also, on inspection, a diagnosis of something other than artificial intelligence.
A Post Already Occupied
Read the requirements as a list and notice what they describe. Selection among rival uses of scarce capital. Explicit go and stop authority. Discipline about time to value. Work made inseparable from how the business creates value. This is not the specification of an emerging technical domain. It is the specification of the strategy function, and it has been understood as such since Joseph Bower demonstrated that strategy in large firms is not formulated at the top and executed below but constituted by the accumulated pattern of resource commitments made through the middle. In other words, what the client of the executive search firm has described is a Chief Strategy Officer and then attached a technology to the title.
The instinct to create a post rather than confront an incumbent is not an operational decision. It is a governance evasion, and it is available precisely because creating a role requires no one to be told anything unwelcome.
The distinction matters because the application of these systems is not a domain problem at all. It is the question of which parts of an existing strategy they make cheaper, faster, or newly possible—and that question cannot be answered from outside the strategy, only from within it. An organisation that appoints someone to run its artificial intelligence has already conceded that the work sits alongside how it creates value rather than inside it, which is precisely the outcome the client said it feared. The office is erected to prevent the very separation the office creates.
We have run this experiment before, and with some regularity. The Chief Digital Officer, the Chief Transformation Officer, the Chief Innovation Officer: each was created because a capability the organisation needed did not exist in its line management, and each was staffed by people whose real mandate was to compensate for that absence rather than remedy it. Such roles are scaffolding. They announce that the building is unfinished, and properly understood they carry a removal date—which is the awkward implication of the client's own ambition. If the work is to become inseparable from how the business creates value, the officer's success condition is the abolition of the office.
There is an honest case for the appointment, and it should be conceded. A dedicated senior post can function as a forcing device: a way of concentrating attention, breaking a stalemate between business units, and giving somebody the standing to refuse a general manager. That is a real use and occasionally the only one available. But it is a temporary expedient, and the intellectually serious version of the brief says so aloud—names the role transitional, sets a horizon, and specifies what must be true in the line organisation before the office closes. A permanent appointment made to solve a temporary coordination failure does not solve it. It institutionalises the failure and gives it a salary band.
Everything Else Is a Side Hustle
Why, then, does the new office get created? Not from stupidity. It is created because the executives already holding the relevant authority have a primary focus, and treat everything outside it as a side concern to be attended to when the main work permits—which is almost never. Particularly in "lean" organisations.
The charitable reading is structural, and it has empirical backing. Attention is the genuinely scarce resource at the top of an organisation, and what a firm does is largely determined by what its decision-makers are situated to notice. Executive job demands are not evenly distributed, and the executive facing severe demands falls back on prior experience and simplified heuristics rather than fresh analysis, which is exactly the failure mode observed by the firm seeking a Chief AI Officer. On this reading, the Chief Strategy Officer—if the role exists—is not neglectful. They are saturated, and a second pair of hands is a rational response.
The less charitable reading is harder to dismiss. An executive who cannot hold a second, third, or fourth serious concern within their remit is not fully performing the office they occupy. Saturation is a real constraint, but it is also the most respectable available account of an incapacity that nobody wishes to name, and organisations reach for it with a consistency that should make us suspicious. The instinct to create a post rather than confront an incumbent is not an operational decision. It is a governance evasion, and it is available precisely because creating a role requires no one to be told anything unwelcome.
It also imposes a cost that the recruiting brief does not price. Accountability is not conserved when it is divided; span of accountability that exceeds span of control produces executives who are answerable for outcomes they cannot command, and the predictable response is defensive rather than entrepreneurial. Insert a new officer between the strategy function and a set of decisions that were formerly the strategy function's, and the question of who is answerable when the portfolio is misallocated has been made materially harder to answer than it was the week before. Two people now hold part of it. Neither holds the whole.
What Cannot Be Specified
Having correctly rejected one proxy—technical credentials standing in for the capacity to create value—the brief reaches immediately for another, and this one is considerably harder to interrogate. Transformational leadership, not a technical build.
Transformational leadership is not a neutral description. It is a specific construct, measured for four decades and criticised for nearly as long, and the criticism has largely converged. The measure cannot reliably separate what the leader does from the results it is supposed to explain, so it risks defining good leadership as whatever preceded a good outcome—including when a leader is a bully who systematically undermines organisational morale. Its component parts, meant to capture distinct qualities, collapse into one another when tested. And its apparent power to predict performance largely evaporates once researchers correct for the fact that the same person is often rating both the leader and the results. For all that, transformation roles survive not because they are well specified but because their existence is flattering and unfalsifiable. Exactly the properties that make a criterion attractive to a selection process. No candidate has ever been rejected for insufficient transformational leadership in terms that are not easily contestable. Resulting in swathes of people being appointed to Chief Transformation roles when their only real talent is driving people harder than before.
The closing formulation compounds the difficulty: a business leader with a natural inclination towards technology, not the other way around. The sequencing instinct is right, and the reverse has produced a great many expensive disappointments. But a natural inclination is a taste, not a capability. It can be performed in interviews by anyone with a subscription to the Harvard Business Review or the Economist and a plausible manner, and it cannot be assessed by any method senior search actually deploys—a field that continues to prefer subjective impression to validated instruments long after the evidence on their relative accuracy ceased to be seriously disputed. As one researcher put it: "people have an inherent resistance to analytical approaches to selection because they fail to view selection as probabilistic and subject to error."
What the client actually wants is phronesis: the capacity to judge particulars, to weigh incommensurable claims, to recognise when a thing that is working is nevertheless not worth continuing. Aristotle's point about this virtue is not that it is rare, though it is. It is that it cannot be reduced to a neat rule, cannot be transferred by instruction, and is exercised only by the person who bears the consequences of exercising it. It does not travel with a job description. It does not, on the available evidence, travel especially well with the individual either—portable excellence being a good deal less portable than the market for executive talent assumes. And it cannot be subcontracted, which was Burke's structural rather than sentimental objection to the representative who surrenders the judgement they were trusted to exercise: the trust attaches to the judgement, and where the judgement is delegated onward, there is nothing left for the trust to hold to.
So the question worth putting to any board contemplating the appointment of a Chief AI Officer is not whether the specification is right, nor whether transformational leadership is the register in which to write it. It is why the executives who already hold the authority to prioritise, to fund, and to stop are not doing so—and whether the vacancy about to be advertised is a remedy for that condition or the most expensive available way of not discussing it.
Good night, and good luck.
