Do trustees need to understand AI to scrutinise an AI project?
No. Every question a board needs to ask can be put and answered in plain business language, and nobody in the room needs to know what a model is.
Blog
What boards should ask before approving an AI project
Six plain questions, none needing technical knowledge: what organisational data goes into it and who approved that; what the organisation would do if the supplier doubled the price or withdrew the product; who can explain in plain terms how it reaches its answers; which decision it will never be allowed to make alone; what success looks like and when you will know; and who will be accountable for it in a year. Agree the answers at approval and record them in the minutes.
What should a board ask about the data and the supplier?
What organisational data goes into the tool, from which systems, put there by whom, and who approved that. In most organisations the honest answer at approval is that nobody has looked, because a team adopted the tool rather than anyone procuring it centrally. For charities and anyone holding data about vulnerable people this matters most and is most likely to be answered vaguely, and vagueness is itself the finding.
Then test dependency. If the supplier doubled the price tomorrow, what would we do? If the answer is that we would pay, the organisation does not control the relationship. The harder version: if the supplier withdrew the product, what would we lose and how long would replacing it take? A tool that has quietly become the way a department works is a bigger commitment than its invoice suggests.
What should a board ask about how the tool decides?
Who in the organisation can explain, in plain terms, what it was trained on, what it is good at and where it is known to be unreliable. If nobody can, the organisation cannot defend a decision the tool influenced, and that becomes a problem the first time an outcome is challenged.
Then ask which decision it will never be allowed to make on its own. In my experience this produces the most useful conversation in the room, because it forces the board to say where human judgement is not negotiable: eligibility, safeguarding, hiring, anything affecting an individual's access to a service. Write the answer down, because a boundary that exists only in people's heads is not a control.
What should a board ask about success and ownership?
What will be different, by when, and what would count as the project not having worked. AI proposals arrive with a lot of enthusiasm, and enthusiasm makes people vague about measurement, so agree the failure condition at approval while everyone is calm. Then ask who will be accountable in a year, once the person who was excited about it has moved on. Adoption without ownership is how organisations end up with tools nobody can explain, switch off or admit to choosing.
Why is AI a board matter rather than a management one?
Because it has three features that make it governance. It moves organisational data outside the organisation, often under terms nobody has read. It influences decisions that affect individuals, which brings duties that sit with the board. And it creates dependencies that are expensive and slow to unwind, usually after the person who approved it has left. Trustees and non-executive directors carry personal accountability for oversight, and "the executive team handled it" is an uncomfortable answer when an AI-influenced decision goes wrong.
What should charity boards watch for?
Enthusiasm about efficiency, usually from someone with something to sell, landing on organisations with little capacity to evaluate it. AI can genuinely help organisations with limited resources do more, and I have seen it do so. Charities also carry obligations that make careless adoption more serious than in a commercial business of the same size: restricted funding cannot be moved to rescue a mistake, trustees are personally accountable, and the data is often about people who are in no position to complain. That argues for the six questions, asked properly and minuted.
Why can a board ask these without technical knowledge?
Because every one is a normal governance question in plain business language. What I notice at board level is less fear of AI than uncertainty about whether trustees are allowed to ask about it. I am a trustee as well as an advisor, so I have asked these questions from inside a board, and I have also presented proposals to boards whose job was to be sceptical. I have no software to sell, no reseller agreements and no platform commissions, so I have no interest in whether a proposal is approved.
If your board wants an independent view on an AI proposal, AI strategy and governance covers oversight and responsible adoption, and an AI readiness assessment answers whether the organisation is ready to adopt anything yet.
Related evidence: Charity coaching. Earlier in this journey: The charity digital spend checklist: questions to answer before you commit a penny.
Questions people ask
Why agree a failure condition at approval?
Because enthusiasm makes people vague about measurement, and nobody agrees what failure looks like after the event. Settle what would count as the project not working while everyone is calm.
Not sure where to start?
Most clients begin with a conversation. No pitch, no hard sell.
Just a straightforward discussion about where you are and whether I can help.
Book a free 30-minute call