AI Governance
Board Questions Every AI Programme Should Answer
A concise set of questions that lets a board test whether an AI programme is under control — without becoming technical.
A board does not govern an AI programme by understanding its technology. It governs by asking the right questions and being unsatisfied with answers that describe activity rather than control. The value of a good question is that it reveals, quickly, whether a programme is genuinely in hand or merely in motion — and it does so without requiring directors to become experts.
The questions below are deliberately plain. Each is designed to surface something a well-run programme can answer with evidence and a poorly run one cannot answer at all. Taken together, they let a board form a reliable view of whether ambition and control are in balance.
On value and purpose
- What specific commercial value is this programme expected to create, and how will we measure it?
- What are we choosing not to do, and why?
- How much are we spending, and against what return?
On accountability and control
- Who is accountable for AI in this organisation, and do they have the authority to enforce the framework?
- How do we know our policy is actually being followed, and how would we detect if it were not?
- Where is AI touching our most sensitive decisions, clients or data, and who owns that risk?
A good question reveals, quickly, whether a programme is genuinely in hand or merely in motion.
On risk and resilience
- What could go wrong, what would the consequences be, and what controls reduce them?
- Could we explain to a client or regulator how AI was used in a given matter?
- What would happen, and who would answer, if an AI-assisted output caused harm?
On judgement
The final question is the one that underlies all the others: are we adopting AI with the same judgement we apply to everything else that matters? A board that can ask these questions, and that insists on answers grounded in evidence rather than reassurance, is exercising exactly the oversight responsible AI adoption requires. Where the answers are thin, that is not a reason to stop — it is the agenda for getting the programme under control.
If this raises a question for your firm, we are always glad to discuss it in confidence.
Book a Confidential Discussion