The brief that most nomination committees give their search firms when looking for an AI-credible non-executive director looks something like this: “We need someone with AI expertise who can advise the board on AI strategy and governance.”
This brief will produce candidates who are technically knowledgeable about AI, possibly from a research or product background, with impressive credentials and an ability to explain transformer architectures to non-technical board members. It will not reliably produce candidates who are effective board-level AI governors.
Technical knowledge of AI and board-level AI governance effectiveness are related but distinct capabilities. The nomination committee that conflates them makes a predictable error, and it is an expensive one to correct once the appointment is made.
The credential that predicts governance effectiveness
The most reliable predictor of board-level AI governance effectiveness is not AI technical knowledge. It is a combination of three things that most AI experts do not have, and that most board-credentialed candidates without AI expertise do have in varying measure.
Governance accountability experience. Has this candidate previously sat on a board, a regulatory body, or a governance committee where they were personally accountable for a domain-level decision — not advised on it, but decided it and bore the consequences? AI governance effectiveness requires understanding what accountability feels like from the inside. Advisors, consultants, and technical experts who have never been in a governance accountability role will typically advise effectively and fail the accountability test when their advice is rejected.
Commercial and operational experience in technology businesses. Has this candidate built, scaled, or operated a technology business — not just researched or advised on one? The governance gap that most AI-deploying companies face is not technical. It is the gap between the technical team’s AI ambition and the board’s ability to evaluate whether the proposed deployment has a credible governance structure. Closing that gap requires someone who has been inside a technology business under operating pressure, not just someone who has studied AI systems academically.
Ability to ask the governance question without getting drawn into the technical answer. This is the rarest combination and the most important. The AI-credible NED who gets drawn into technical discussions — who is most valuable in the meeting when the architecture is being discussed — is not functioning as a governance body member. They are functioning as a technical advisor to the board. The AI-credible NED who can consistently re-route technical discussions back to the governance question — “I understand the architecture; I want to understand the oversight mechanism” — is providing board governance, not technical expertise.
The profile that fails most often
The most common appointment failure I see in the AI NED space is the academic or research-background AI expert who has never operated a business or sat on a governance body.
The failure mode is specific: they are excellent in the meetings where AI strategy is being discussed, and largely absent from their governance function in the meetings where procurement, employment, or financial systems happen to involve AI. The AI that poses the highest governance risk to most mid-sized companies is not the flagship AI strategy project — it is the HR system’s AI-driven performance ranking, the accounts payable automation, the customer service chatbot that is making eligibility decisions. These are Annex III risk areas that a board member without operational experience may not recognise as governance obligations.
The research-background AI expert also tends to evaluate AI proposals on technical quality rather than governance quality. A well-designed AI deployment with inadequate human oversight will typically receive less scrutiny than a less elegant deployment with robust governance — because the technical quality is legible to the expert and the governance quality requires different assessment criteria.
The interview questions that reveal what you need to know
Three questions that reveal the capability behind the credential:
“Walk me through the last time you disagreed with a board decision you were part of. What did you do?” The answer to this question reveals: whether the candidate has actually been in a governance accountability role, what their approach to disagreement is (accommodation, escalation, resignation — each has different governance implications), and whether they understand the difference between advising and governing.
“Describe a situation where an AI system you were involved with produced an unexpected output. What was the governance response?” The answer reveals: whether the candidate has operational AI governance experience or only design-stage experience, what they consider an adequate governance response, and whether their instinct is to describe the technical cause or the governance failure.
“Our CTO is proposing an AI deployment that would fall under EU AI Act Annex III high-risk classification. The deployment would materially improve our customer service response times. What governance questions would you want answered before the board approves it?” The answer reveals: whether they can translate AI governance knowledge into the specific governance questions a board needs to answer, whether they know the EU AI Act’s board-level obligations, and whether their first instinct is to evaluate the technology or to establish the governance architecture.
What to prioritise in the shortlist
If the nomination committee has produced a shortlist of candidates with genuine AI knowledge, the selection criterion should be governance disposition, not technical depth.
The candidate who, in a board setting, will consistently re-route technical discussions to governance questions is more valuable than the candidate who can answer the technical questions. The board has a CTO, a CISO, and a technical team. What it does not have — or is looking to add — is a governance voice for AI that is credible enough to be taken seriously by the technical team while maintaining the board-level perspective on accountability, risk, and oversight.
That combination is rarer than technical AI expertise, and it is harder to find through a standard search firm brief. The brief needs to be written around governance disposition, not technical credential.
The Board AI Governance Framework provides the governance structures that an AI-credible NED appointment needs to operate effectively once in post. It also defines the oversight and approval criteria that the nomination committee can use to assess whether a candidate’s governance approach is aligned with what the board actually needs.
For boards seeking independent advisory support on AI governance and technology board composition, contact Steven directly.
