In the AI board meetings I have sat in, not one question was asked about the technology. The agenda was budget, then who would run it, then the dates of the next presentations. That is not negligence. It is what happens when experienced directors are asked to govern something they cannot afford to admit they do not understand.
What the meeting is actually about
The pattern repeats with unusual consistency.
Budget comes first. Then the political question of who leads it, because the person given an AI project is seen as a coming figure inside the company and outside it. Then toll gates and the dates on which the project will report back, because the board enjoys the power to let something continue or not.
On one project of my own, the entire meeting was about staffing. Board members pushed their preferred people onto the team, not because of what those people knew, but because of the visibility the project carried.
Nobody asked what the system would actually do, what data it needed, or how anyone would know whether it had worked.
Why nobody asks a real question
This is the mechanism, and it is worth stating plainly because almost nobody writes it down.
A board member with decades of experience, expert in their own field, responsible for thousands of people, cannot sit in a room and admit they know nothing about AI. So they ask a question in a professional tone. It sounds appropriate. It ticks the box. I asked a question. I did not look foolish.
What they will not do is ask a real one. Because a real question produces an answer wrapped in acronyms and jargon, and then they will not know whether the answer was any good. Not knowing in private is uncomfortable. Not knowing in front of the board, on the record, is worse.
Meanwhile the person presenting leaves thinking they handled the boardroom well. Everyone is satisfied. And the project has been done a serious injustice, because the one forum with the authority to improve it never engaged with it.
This is not stupidity. It is people defaulting to what they know when the alternative is exposure. The pattern is measurable: Protiviti’s 2026 global board survey found that only around a quarter of boards discuss AI at every meeting, while among organisations reporting strong AI returns the figure is more than double that.
The fix costs nothing
Introduce the board to AI before the presentation, not during it.
An hour added to a board meeting. A background session with no decision attached. A briefing some months ahead of the first proposal, so that directors have time to become comfortable somewhere other than in the room where they are being asked to approve something. The discomfort is the problem, and time is the only thing that removes it.
There is no list of AI questions, and that is the point
Directors often ask me for the list of questions to put to an AI project. There isn’t one, because it depends entirely on the project. But the ordinary governance questions work perfectly well, and none of them require any AI knowledge at all.
If this were not an AI project, how would you run it? Does it meet our normal standards for time, budget and reporting? Do you have the resources, the mandate and the support you need? What is missing that you need in order to succeed?
Those are the questions a board already knows how to ask. AI does not suspend them. If you want the ones that come the other way, the three your board is most likely to put to you are a separate and sharper problem.
The one question that matters most
And then there is the question that requires no technical knowledge whatsoever, and which changes the temperature of the room.
Are you personally accountable for the outcome of this project if it does not succeed? Are you willing to carry that, given this is one of our first AI projects?
Ask it seriously and you will learn more in the next thirty seconds than in the previous hour. If the answer is yes, you have a real owner. If it is hedged, you have discovered something important before you spent the money rather than after.
Boards spend a great deal of energy trying to become knowledgeable enough to evaluate AI. Accountability is a faster instrument, and it is one they already know how to use. It is also the question most likely to be asked of them in turn.
Questions this article answers
Why does nobody ask about the technology in an AI board meeting?
Because the agenda defaults to what directors already know how to govern: budget, staffing and reporting dates. A real question about the technology risks an answer nobody in the room can evaluate, so it is never asked, and the silence is mistaken for agreement.
Why do boards fail to scrutinise AI projects?
Because directors cannot comfortably admit they do not understand the subject, and a real question risks an answer they cannot evaluate. A professional-sounding question that ticks the box feels safer than exposure.
Does the board need AI training before approving a project?
It needs exposure, not expertise. A background session or a briefing well before any proposal gives directors time to become comfortable outside the room where they must decide.
What is the single best question to ask an AI project sponsor?
Whether they are personally accountable for the outcome if it does not succeed, and whether they are willing to carry that. It requires no technical knowledge and reveals whether there is a real owner.
Is it the board’s fault when an AI project fails?
Responsibility is shared, but the board is the one forum with authority to improve a project before it starts. When it does not engage, that opportunity is lost.
For more on this, see Strategic AI Advisor Meetings, part of the Strategic AI Advisor guide.
If you are leading your organisation through this, this is the kind of work I do with a limited number of senior leaders each quarter. → https://anglero.com/the-reset/
Strategic AI Advisor, MerkabaPhi AS, Oslo. 450+ keynotes across 30+ countries.
Listen to the Clarity at the Top podcast on Spotify.
Enquiries: anglero.com