There is a hiring problem starting to show up in boardrooms. AI becomes a priority, a steering group is formed, someone is appointed Chief AI Officer, Head of AI or Head of Transformation, tools are bought, pilots are launched, and the new leader is asked to “drive adoption across the business”. On paper, it all looks sensible. Six months later, the same questions are still being asked.
Why is AI usage so patchy?
Why are there so many disconnected pilots?
Where is the commercial return?
And who actually owns the outcome?
Quite often, the problem goes all the way back to the original brief.
McKinsey’s 2026 global survey gives some useful context. Eighty per cent of respondents say AI has improved their own productivity, yet only 37% report a positive impact on EBIT, unchanged from last year. Just 6% qualify as AI high performers. There are plenty of reasons for that gap, including technology, data, governance, skills and change management, but there is also a leadership question that I think gets overlooked: who is actually responsible for turning AI into a better-performing part of the business?
In many organisations, the answer is surprisingly unclear.
The impossible AI brief
I’m seeing AI leadership roles asking candidates to do almost everything at once: understand the technology, own the data agenda, influence the board, redesign processes, get people using the tools, change the culture, manage risk, find efficiencies and create new revenue.
It is a huge brief for one person, and often a sign that the executive team has not yet worked out where the real accountability sits.
Is this person setting the company-wide AI strategy, or changing how a particular part of the business works?
Can they actually make decisions about technology and data, or are they there to advise?
Do they own a commercial result?
Can they change a workflow, stop a pilot, remove a tool or challenge a senior functional leader who does not want their area disrupted?
If the leadership team cannot answer the above questions, the search very quickly becomes messy. The brief changes, strong candidates somehow feel “not quite right”, and different stakeholders have different versions of what the role is there to do. Recruitment can expose that problem, but it cannot resolve it on behalf of the executive team.
So who should companies be looking for?
McKinsey also studied 20 companies that have managed to create sustained value from technology and AI, and one thing stood out. They concentrated on a small number of important areas of the business and went deep. They connected the technology to a real performance problem, then changed processes, data, incentives and ways of working around it.
Crucially, there was a senior business leader accountable for making it work. That person understood the business inside out, knew enough about data and technology to make good decisions, and had the authority to get things changed. In some cases they worked closely with a senior technology leader, but accountability for the business outcome was still clear.
I think this is one of the leadership hires many organisations are missing. The job title could be almost anything. What matters is that this person owns an important part of the business where AI could genuinely make a difference, whether that is customer acquisition, product development, operations, supply chain, client service, research or commercial planning.
They also need enough authority to change how that part of the business actually works. That means being able to make decisions about people, process, technology, data and investment, have difficult conversations with established leaders, and recognise the difference between an impressive demo and something that will genuinely improve performance.
For insights businesses, that might mean someone who can rethink how research is produced, how evidence reaches decision-makers, how expertise gets turned into products, or how client work is delivered. That is a very different search from simply looking for somebody with “AI” in their job title.
Which makes the talent market much more interesting
The people capable of doing this will not necessarily have obvious CVs. Some will be operators who have transformed their own function and picked up serious technology experience along the way. Others will come from product, data or strategy and have developed real commercial responsibility. Some will have been through major transformation programmes and know first hand how difficult it is to change an organisation once the PowerPoint is finished.
What I would want to understand is what they have actually changed. What did they stop or simplify? Which processes did they redesign? Where did they meet resistance? What decisions were they personally able to make? And what happened commercially as a result?
That is where the evidence is. It also means that searching purely by a job title will miss a lot of the market because some of the strongest candidates may never have held an AI title at all. The search has to start with the business problem and the authority you are prepared to give the person. Once those are clear, you have a much better chance of finding the right people.
Three things I would want a C-suite to decide before opening the search
1) Where do you want AI to make a meaningful difference first? Be specific about the part of the business, the outcome you want to improve, the workflow that needs to change, the data sitting behind it and, importantly, what this person will actually be allowed to decide.
2) Are you really looking for one person? In some businesses, pairing a strong business operator with a strong technology leader will make far more sense than searching endlessly for someone who can do everything. That only works, though, if both people know where accountability sits.
3) Hire for evidence, not just AI fluency. Most senior candidates can now talk convincingly about AI. I would be much more interested in what they have actually changed: what they stopped, what they consolidated, which decisions they moved, how they dealt with resistance, and what happened to revenue, cost, speed, quality or customer experience afterwards.
Only 6% of organisations in McKinsey’s research are currently classed as AI high performers. Looking at what separates them, I keep coming back to leadership. They have made choices about where AI matters, who owns the result and what authority that person has to change the business.
For any CEO thinking about their next AI leadership hire, those decisions probably need to come before the job description.
The Job Description Is a Management Document
Most people think a job description is a recruitment document. It is more revealing than that.
For a senior AI or transformation appointment, it is one of the clearest documents an organisation produces about what it believes is broken, what it wants to protect, and whether its leadership team has agreed how the future will work.
A badly written brief gives the game away. It asks for “strategic vision and hands-on execution”, “deep technical expertise and outstanding stakeholder management”, “a transformation mindset and proven delivery”, “board-level influence and the ability to roll up sleeves” (the last one is a particular favourite of mine).
There is nothing wrong with any of those qualities. The problem is that, placed together without a clear mandate, they become a shopping list for a problem the organisation has not defined.
The result is predictable. Candidates are interviewed against different private expectations. The technology leader wants someone who can solve the architecture. The CEO wants someone who can make the business move faster. The HR leader wants a visible culture carrier. The functional heads want somebody who will help without changing too much. Nobody is wrong, but the role becomes impossible to fill because it is designed to absorb disagreement rather than resolve it.
The talent market then gets blamed for a shortage of “AI leaders”. In reality, very few people can succeed with a HUGE mandate and very little authority. There are plenty of capable executives who can transform a clearly defined business problem when they have the backing and authority to do it.
McKinsey’s latest survey reinforces the point. Organisations that create significant value use AI to pursue growth and innovation as well as efficiency, and nearly three quarters of its high performers have fundamentally redesigned workflows. Only one quarter of other respondents report the same.
That scale of change cannot sit inside a vague job description. A serious leadership brief has to say where growth will come from, where cost will move, what workflow is in scope, whose team changes, what data the leader can access, and how success will be measured.
The best briefs I see now do four things.
They identify one business problem worth solving rather than announcing an enterprise-wide ambition. They spell out what the person can actually decide. They make it clear how the role works with technology, data, HR, legal, finance and the CEO. And they define success through outcomes, not activity.
For an insights team, for example, that might mean a leader who owns the redesign of the client evidence to recommendation workflow. Their measures could include turnaround time, quality of synthesis, strength of source traceability, commercial conversion, client confidence, and the development of junior talent. The brief would state what they can change and which decisions still require another executive’s approval.
That is a ‘recruitable job’. More importantly, it is a manageable one.
The smartest question a candidate can ask during a senior AI search may be the simplest: “What am I genuinely accountable for?” If the answer takes ten minutes and ends with “it depends”, the organisation has work to do before it hires anybody!
This piece first appeared in Talent Pools Nexus, our weekly read for leaders in research, insights and consulting. Subscribe on LinkedIn.
