The Five Questions Your Board Should Ask Before the Next AI Budget

Most AI budgets get approved on a mix of faith, fear of missing out, and an impressive demo. A year later, nobody can quite say what happened to the money. The fix is not more oversight. It is five sharp questions, asked before the budget is approved, not after it is spent. Any CEO should welcome them, and any program that cannot answer them is not ready.

The five questions

1. What number does this move, and in what unit? If the answer is "efficiency" or "innovation," it is not a business case. Demand dollars, hours, risk, revenue, or cycle time. A number you did not name is a number you cannot defend.
2. Are we building an asset we own, or renting a capability everyone rents? If the advantage comes from a general model on a generic workflow, every competitor buys the same thing. Ask what here is ours, the data, the workflow, the judgment, that a rival cannot buy tomorrow.
3. What is the plan for year two? AI value compounds in the second year, through the platform, the data flywheel, and the eval loop. A program funded for a quarter and expected to pay off for a decade is mis-designed. Ask what makes this compound, not just launch.
4. Who owns it when the pilot ends? Most AI dies in the gap between a successful pilot and a production owner. Ask which team runs this, on which platform, with what budget, after the excitement fades. If nobody owns year two, there will not be one.
5. What could go wrong, and who is accountable? Governance, data, model error, vendor lock-in. Ask what the failure modes are and who carries the risk. An AI program without a named owner of its downside is an incident waiting to be scheduled.

Why these five

Each question kills a specific way AI programs waste money: unnamed outcomes, rented advantage, quarter-thinking, orphaned pilots, and unowned risk. A program that answers all five is not guaranteed to succeed, but it is honest, and it is fundable. A program that cannot answer them will absorb budget and produce slides.

So before the next AI budget lands on the table: which of these five can your program not yet answer?