What Boards Get Wrong When They Ask for an AI Strategy

The Wrong Question Gets the Wrong Answer

There is a specific board question that I hear in every enterprise AI conversation: "What is our AI strategy?" The question seems reasonable. It is, in fact, the wrong frame — and it produces the wrong answer almost every time.

When the question is "what is our AI strategy?", the answers that come back are technology-centred: which models to evaluate, which vendors to pilot, which use cases to explore. These answers are not wrong. They are just several steps removed from the decisions that actually determine whether an organisation gets durable value from AI investment.

The Question That Gets the Right Answer

The question that produces useful answers is: "Which of our highest-value business problems are currently bottlenecked by human judgment at scale?" This reframe shifts the conversation from capability to constraint. Instead of asking what AI can do, you are asking where your organisation's decision-making is the limiting factor on performance — and whether that constraint is solvable with the right AI application.

The difference sounds semantic. The consequences are not. A technology-first question produces a list of pilots. A constraint-first question produces a prioritised backlog of operational problems with dollar values attached, owners named, and success metrics defined before any model is selected.

Why Boards Default to the Technology Question

Boards ask about AI strategy because they are responding to external pressure — from investors, from competitors, from media coverage. The question "what is our AI strategy?" is the right question to ask a technology vendor. It is not the right question to ask an executive team, because it implicitly assigns the executive team the role of the technology vendor: evaluate the landscape, choose a platform, run pilots.

Executive teams have one job with respect to AI, and it is not technology selection. It is identifying and prioritising the operational problems where AI investment will produce measurable business return — and building the organisational capability to deliver on that priority. Technology selection follows from problem definition. When it precedes problem definition, it produces expensive pilots that do not convert to operations.

What a Useful Board Conversation Looks Like

A useful board-level AI conversation has three parts. First: which business problems, ranked by cost and decision volume, are we trying to solve? Second: for the highest-priority problem, what does operational success look like in measurable terms — not "AI adoption" but a specific number in a specific time frame? Third: who in the organisation owns the outcome, and does that person have the authority and the budget to see it through?

These are governance questions, not technology questions. They are appropriate for a board. The technology question — which model, which vendor, which architecture — is appropriate for an engineering or product team once the governance questions are answered.

The Accountability Gap

The most common failure mode in board-level AI governance is not the wrong technology choice. It is the absence of a named owner with a P&L accountability for the outcome. When AI investment is managed as an IT programme or a research function, the board has no way to hold anyone accountable for business results, because the people running the programme are not accountable for business results — they are accountable for deploying technology.

The governance intervention that produces the highest return on board time is simple: for every major AI initiative, require a named business owner who reports to a P&L and who has committed to a measurable business outcome with a deadline. The technology team reports to that owner. The board monitors the business metric, not the technology programme.

A Note on Sequencing

None of this means boards should not ask technology questions. It means the technology questions should come second. "Given our priority problem and our success metric and our named owner — which technology approach gives us the best path to that outcome?" is a question that produces useful answers. It is also a question that the people who will be held accountable for the outcome are well-positioned to answer.

The board's job is to ensure the first set of questions is answered well. If it is, the second set largely answers itself.