A board does not need to choose a model or configure a tool. It does need enough clarity to decide whether an AI initiative fits the organisation’s strategy, risk appetite and obligations.
1. What decision or outcome is this meant to improve?
“Use AI” is not an outcome. A proposal should identify the customer, operational, revenue, quality or risk problem it intends to address—and how success will be measured.
2. Why is this the right use case now?
Ask what evidence supports the priority, what alternative approaches were considered and what would make the initiative not worth pursuing.
3. What changes for people?
AI affects roles, skills, workflows and trust. Management should explain where human judgement remains essential and how staff will be supported to use the system well.
4. What data and supplier dependencies are being created?
The board should understand the data involved, the security and confidentiality implications, and what happens if a supplier changes terms, performance or availability.
5. Who owns the risks and the decisions?
Responsibility cannot sit with “the AI team” alone. Name the executive owner, operational owner and the route by which significant issues reach the board.
6. How will we know whether to scale, change or stop?
Set review points in advance. A good pilot produces a decision, not simply another pilot.