The C-Suite AI question nobody is asking.

Every board is talking about AI. Most are asking the wrong question. The question that will actually determine whether AI delivers commercially isn't about the technology — it's about the organisation. And almost no one is asking it.

In the past two years, a familiar pattern has emerged across organisations of every size and sector. The board commissions an AI strategy. A working group is established. Consultants are engaged. A roadmap is produced. Investment is approved. Tools are selected and deployed.

And then, somewhere between the strategy and the reality, progress stalls. Adoption is slower than expected. Teams are using the tools differently to how they were designed to be used — or not using them at all. The productivity gains that seemed straightforward in the business case are proving elusive in practice. The technology worked. The organisation didn't.

This is not a failure of AI. It is a failure of organisational readiness — and it was predictable from the start, because the question that would have identified the risk was never asked.

The question most boards ask

"What can AI do for us?" is a reasonable starting point. It generates useful answers about automation, efficiency, capability augmentation and competitive advantage. Technology vendors answer it fluently. Strategy consultants have well-developed frameworks for it. It fills a boardroom agenda item comfortably.

But it is the wrong question to lead with — because it focuses attention on the technology rather than on the organisation that will need to absorb, govern and benefit from it.

The technology question is easy. The organisational question is hard. Which is probably why most boards spend most of their time on the wrong one.

McKinsey's research on large-scale technology transformations (2023) found that less than 30% of AI and digital transformations fully deliver on their stated objectives. The primary reason cited was not technical failure but organisational factors — culture, capability, leadership behaviours and change management. The pattern is almost identical to what the same firm found about broader change programmes in 1996 and again in 2009. The technology changes. The human constraint stays remarkably consistent.

McKinsey & Company (2023) 'The State of AI in 2023'; McKinsey & Company (1996, 2009) 'Organisational Change' research series.

The question nobody is asking

"How do we lead an organisation that's learning to use AI?" is the question that actually matters — and it is almost entirely absent from board-level AI conversations.

It is a harder question. It doesn't have a clean answer that can be packaged into a strategy document or a vendor roadmap. It requires leaders to examine their own behaviour, their organisation's culture, and the operating model conditions that will either support or undermine adoption. It invites uncomfortable scrutiny of things boards typically prefer to leave unexamined.

But it is the question that determines outcomes. Because the organisations that successfully integrate AI are not distinguished by having chosen better technology — they are distinguished by having created the conditions in which people could use that technology effectively. That is a leadership and organisational design challenge, not a technical one.

What the research tells us about technology adoption

The behavioural science of technology adoption is well-established. B.J. Fogg's Behaviour Model (2009) identifies three elements that must be present simultaneously for behaviour change to occur: motivation, ability, and a prompt. Most AI implementations invest heavily in the prompt — announcements, training programmes, mandatory usage policies — while underinvesting in the ability (genuine capability to use the tool confidently) and almost entirely ignoring the motivation question (why would someone choose to use this when their existing approach already works well enough?).

Fogg, B.J. (2009) 'A Behavior Model for Persuasive Design', Proceedings of the 4th International Conference on Persuasive Technology.

Everett Rogers' Diffusion of Innovations research (1962, updated 2003) identified that technology adoption in organisations follows a predictable curve — with innovators and early adopters typically representing around 16% of a population, and the majority requiring social proof, reduced perceived risk, and visible endorsement from credible peers before changing their behaviour. Most AI implementations are designed as if the entire organisation is composed of early adopters. It isn't.

Rogers, E.M. (2003) 'Diffusion of Innovations', 5th edition. Free Press.

Rogers, E.M. (2003) 'Diffusion of Innovations', 5th edition. Free Press.

The implication is clear. A successful AI implementation requires different interventions for different groups — and the majority require something that most AI programmes don't provide: visible, credible leadership behaviour modelling the change. When the C-Suite is seen to use AI tools themselves, to talk about how they're using them, to normalise the learning curve and the occasional failure — adoption accelerates. When the C-Suite endorses AI in presentations but doesn't visibly use it themselves, the majority reads the signal correctly and waits.

The four questions a board should actually be asking

Before any AI strategy can succeed, four organisational questions need honest answers. These are not questions for a technology workstream. They are questions for the board.

Edmondson, A.C. (1999) 'Psychological Safety and Learning Behavior in Work Teams', Administrative Science Quarterly, 44(2), pp. 350–383.

What this means in practice

The organisations that are getting AI right are not doing so because they chose better technology or hired better data scientists. They are doing so because their leaders treated AI adoption as an organisational change challenge from the outset — not a technology deployment.

That means starting with the human and cultural conditions, not the tool selection. It means investing in the leadership behaviours that create the conditions for adoption before the rollout begins. It means designing for the majority — who need social proof, reduced risk and visible endorsement — not just for the early adopters who would have adopted anyway. And it means measuring outcomes: whether people are using AI in ways that change how they work, not whether they attended the training.

The technology question has a thousand consultants answering it. The organisational question has almost none. That imbalance is exactly why so many AI programmes underdeliver. — Stephen Dixon-Mould FCIPD

The C-Suite AI question nobody is asking is not a technology question. It is a leadership question. And until boards start treating it as one, the gap between AI investment and AI return will remain stubbornly wide.

Three things a board can do this week

1. Audit your own behaviour. Not your AI strategy — your personal behaviour. Are you using AI tools yourself? Visibly? Are you talking about the learning curve, including the failures? If not, the signal you're sending is louder than any strategy document.

2. Ask the psychological safety question honestly. In your organisation, right now — would someone feel comfortable saying "I don't understand how to use this effectively" or "I tried it and it didn't work well"? If the honest answer is no, that's the first thing to fix.

3. Map the operating model friction. Where are the structural conditions that make AI adoption harder than it needs to be? Governance processes that weren't designed for AI outputs. Accountability frameworks that penalise the experimentation that adoption requires. These don't fix themselves through better communication. They need deliberate redesign.

If any of these questions resonate with a challenge you're navigating, start a conversation with Stephen. The first conversation is diagnostic, not commercial.

References: McKinsey & Company (2023) 'The State of AI in 2023'; McKinsey & Company (1996, 2009) Organisational Change research series; Fogg, B.J. (2009) 'A Behavior Model for Persuasive Design'; Rogers, E.M. (2003) 'Diffusion of Innovations', 5th edition, Free Press; Edmondson, A.C. (1999) 'Psychological Safety and Learning Behavior in Work Teams', Administrative Science Quarterly, 44(2). Academic grounding informed by current MSc in Organisational & Business Psychology, University of Wolverhampton.