A market does not simultaneously rush to create a role and doubt its future unless something is unresolved at the role’s core.
Bloomberg reported in June that banks from Sydney to London are racing to fill the role of chief AI officer — a job that barely existed a year ago. The same article carried a quieter finding: the people already in the seat suspect it may not be around for long. One of them, formerly of a major international bank, put it as starkly as it can be put — that anyone taking the role should assume it has no place in the organisation’s future. That is not an outsider’s scepticism. It is the incumbent’s own view of the job.
Hold those two facts together, because they are not a contradiction. They are a diagnosis. And what is unresolved is not whether institutions need senior AI leadership — the regulators of two major markets have just made boards explicitly accountable for AI, and the capability gap beneath that accountability is wide. What is unresolved is what the role actually is. The title has spread faster than the mandate. And a title without a settled mandate churns — which is precisely what the incumbents are sensing.
The mandate problem
Having run transformations for thirty years and built AI systems with my own hands for the last decade, I have watched institutions reach for four different answers to the ownership question, each with a real logic and a real limit.
Some vest it in the chief executive — which honours the truth that AI is a business-architecture question, but founders on the fact that no chief executive can run it day to day. Some vest it in technology — which brings genuine depth, but places the business-model decisions the work requires outside the owner’s remit, so the effort drifts toward automating what already exists. Some build a federated model — business and technology co-owning, a transformation lead threading them — which matches the horizontal shape of the problem but can decay into a programme office with responsibility for everything and authority over nothing.
And some appoint a chief AI officer. Here the language turns slippery, because the market has begun applying that title to the federated model’s coordinator. When the title means that — convener of councils, harmoniser of pilots, publisher of policy — the incumbents’ doubt is well founded. A coordination role is scaffolding. When the organisation learns to hold the load itself, scaffolding comes down. The role that “might not be around for long” is that role, and its disappearance would be no loss.
The role, defined properly
There is a version of the chief AI officer worth building, and it is different in kind, not degree. Picture the role as a seeding ground: a senior owner who works across the existing businesses to transform them — in close conjunction with the business and technology leaders, never around them — and who also carries a mandate the coordinator never gets: to build the institution’s genuinely new, AI-native business, with named revenue objectives and the authority to build from the ground up.
That single addition changes everything about the seat. It turns the role from integrative to future-building. It gives it a P&L trajectory rather than a programme plan. It answers the churn question, because a builder’s role does not dissolve when the organisation matures — it graduates. And it changes who you would hire: not a policy author or an evangelist, but an operator who has run businesses, can challenge the technologists on their own ground, and has built enough with their own hands to know what the technology will and will not bear.
I will be candid about the state of the evidence, because the discipline matters: this early, I cannot point to a named institution running precisely this builder’s version of the role with a documented result. What I can point to is the pattern behind the institutions where AI has actually paid — the work was business-led, horizontally owned, and aimed at revenue rather than deployment counts. The builder’s mandate is that pattern, given a seat at the table.
Five questions before you hire one
For a board about to create this role — or a search partner about to define one for a client — the difference between the coordinator and the builder can be established before a single candidate is met. Five questions do it.
Is the mandate to coordinate what exists, or to build what does not? If the role description is councils, frameworks, and enablement, you are hiring scaffolding. Does the role carry revenue accountability — a named AI-native business objective, not an adoption metric? Does it report to the chief executive, with standing access to the board that now bears regulatory accountability for AI? Could the person you are considering build the thing themselves if they had to — not manage the builders, but genuinely contest the technical choices? And what happens to the role in five years if it succeeds? For the coordinator, success is obsolescence. For the builder, success is a business — and its leader.
That last question is also the answer to the talent problem the churn creates. The best operators will not leave a P&L for a role the market whispers is temporary. Define the seat as a builder’s mandate and the career logic inverts: the chief AI officer who builds the AI-native business becomes its chief executive. The role stops being a stop on the way down and becomes a route to the top — and the calibre of person it attracts changes accordingly.
What the churn is really telling us
The doubt inside the role today is not a verdict on AI leadership. It is the market correcting a definition. Institutions reached for a title before settling the mandate, and the incumbents are living inside that unsettledness. The correction is available to any board willing to make it: own AI where its consequences land, thread it horizontally through business and technology — and give one senior builder the mandate to create what the institution must become, not merely to coordinate what it already is.
The racing demand is right. The doubt is right too. The role that deserves to exist is the one that resolves them both.
