Good governance takes shape in the way a board works: what receives attention, whose views are heard, how assumptions are challenged and what happens after a decision. The board chair has a particular responsibility for making these processes work.
My starting point is that the board chair creates the conditions for the board to exercise sound judgement together. That requires preparation, openness and the willingness to address difficult questions. In the AI era, this has become even more important than before.
The board chair creates the conditions for the board to exercise sound judgement together.
The distinction between leading the board and exercising its powers matters. The board has collective responsibilities for strategic direction and management oversight. The chair helps it fulfil those responsibilities, within the organization’s governance framework. The G20/OECD Principles of Corporate Governance provide a useful foundation for this division of responsibilities.
Working with the CEO, the chair shapes an agenda that gives sufficient attention to the organization’s purpose, strategy and significant decisions. Board members need timely, understandable information and enough time to examine the choices before them.
For AI, I would start with purpose. What are we trying to improve? Who should benefit? How will we know whether the investment is worthwhile? A proposal to introduce AI should make those answers visible. A compelling demonstration may open a discussion, but a board decision needs evidence about the intended use, its limitations and its consequences.
Your board may contain considerable expertise, yet still fail to use it. Some members speak readily; others need space or encouragement. The chair’s approach determines whether questions are welcomed and uncertainty can be acknowledged.
Differences in technical confidence can distort participation.
Seeking consensus is valuable, but disagreement also provides information. Before bringing a discussion to a close, the chair should test whether concerns have been answered and whether apparent agreement reflects a shared understanding. Decisions, material reasons, declared conflicts and dissent should be recorded appropriately.
AI adds a particular challenge: differences in technical confidence can distort participation. A member who asks how an AI-supported decision could be challenged may raise a more consequential issue than the person most familiar with the technology. The chair should make room for both perspectives.
The chair and CEO need a relationship that supports candid conversations, preparation and constructive challenge.
The chair leads the board and the CEO leads the company. They both hold key roles in the organization. The chair and CEO need a relationship that supports candid conversations, preparation and constructive challenge. The rest of the board must also have the information and opportunity to exercise its judgement.
For an AI proposal, the chair can help the CEO understand what the board needs to assess: the expected benefit, alternatives, costs, affected people and safeguards. Management should then bring forward a proposal with clear operational ownership. Decisions requiring board approval belong with the board.
AI governance needs identifiable responsibility.
NIST’s voluntary AI Risk Management Framework emphasises clear responsibilities, communication and executive accountability for AI risks. For a chair, a practical application is to make sure the board can answer a few concrete questions:
- Which AI uses are significant enough to require board attention?
- Who is accountable for each of those uses?
- What evidence supports the expected benefits and acceptable performance?
- What happens when the system produces an error or someone challenges its output?
- Who can suspend its use, and what would trigger that decision?
Consider a collective management organization exploring AI to help match usage records to rightsholders. The board discussion should examine whose works might remain unmatched, how errors could affect distributions and how rightsholders could seek correction. Those questions connect the technology directly to the organization’s purpose.
The chair should also bring the board’s own use of AI into the discussion. AI-generated summaries of board papers, for example, call for agreed practices on confidentiality, checking accuracy and consulting the original material.
The chair should also bring the board’s own use of AI into the discussion.
An approval should specify what happens next: who acts, what conditions apply and when the board will review the outcome. For a significant AI initiative, that review should cover benefits achieved, errors, complaints and material changes in how the system is used. A pilot may justify a limited trial while leaving wider deployment subject to further evidence.
The chair also helps relevant stakeholder concerns reach the board. Listening becomes useful when it influences the questions asked and the decisions taken.
Finally, chairs should apply the same scrutiny to the board itself. Does it have enough understanding to assess AI proposals? Can members challenge expert advice? Are agreed actions followed through?
The quality of chairing becomes visible in the quality of the board’s judgement.
The quality of chairing becomes visible in the quality of the board’s judgement. As AI changes the choices organizations face, the chair’s contribution is to keep those choices connected to purpose, evidence and clear accountability that continues also after approval.
AI will test how well boards govern: whether they ask difficult questions, recognize the limits of their knowledge and follow through on their decisions. The chair has a central role in making that scrutiny possible and expected.
The test comes when the board is asked to approve a promising proposal with unresolved questions. Can it explain why it is proceeding, whose interests are at stake and what would cause it to change course? A strong chair makes sure those questions are answered before approval and revisited when the evidence changes.