Are boards asking the wrong questions about AI?

Boards are simultaneously confronting AI inside the boardroom and AI outside it, reshaping the businesses they govern. The danger is becoming so preoccupied with the first that we under-govern the second.

There is no shortage of conversations about AI. What I find more interesting is how quickly the conversation changes when you put a group of experienced board members together and ask a slightly different question:

That was the subject of our August Competent Boards Global Forum. As always, the discussion was held under the Chatham House Rule, so I will not attribute individual comments. But I can share the collective picture, because it left me with a lot to think about.

And perhaps two thoughts in particular: We want more from AI. We are also worried about what it might take away.

We asked participants where a completely trusted AI tool could add the most value to their board work.

Exploring alternative scenarios came through most strongly, followed by looking ahead for emerging risks and opportunities and testing management assumptions. Understanding consequences and trade-offs was another important area where participants saw real potential.

Think about what those answers have in common. They are not primarily about making the boardroom work faster. They are about making the board's field of vision wider.

What else could happen? What are we missing? Which assumption are we accepting too easily? What connects this issue to something in another paper, another market or another part of the business?

AI can potentially help a director ask better questions. But it also creates new ones: Can we trust the information in our board papers? Has the underlying data been verified? Has AI materially shaped the analysis or recommendation? Could an apparent fact actually be an AI-generated error or hallucination?

This is where the distinction between leading with AI and being led by AI becomes important. I hope every board chooses to lead with AI, not being led by it! But leading with AI demands considerably more than simply having access to it, and considerably more than accepting the information it produces.

For years, we have talked about the T-shaped director: broad enough to contribute across the board agenda, while bringing deep expertise in one or more areas. AI potentially changes that equation.

A board member can now reach into subjects, perspectives and bodies of knowledge that previously would have taken far more time to access. The scarce resource may therefore become less about access to knowledge and more about the ability to interrogate, connect and apply it.

But there is an important catch: Access to more intelligence does not automatically produce better judgement.

In fact, our Global Forum participants were very clear about what worried them. Loss of independent thinking and bias in the analysis came through most strongly, alongside concerns about over-reliance, inaccurate or misleading outputs, confidentiality and security, and the false confidence that can come from appearing better informed. There was also concern about directors using AI without sufficiently understanding its limitations.

What stood out to me was what participants believed becomes more valuable as AI gets better: judgement, ethical reasoning, curiosity, critical thinking and the ability to challenge constructively.

I would add sense-making, and the courage to question an apparently authoritative AI-supported conclusion when experience, context or judgement tells you something does not add up.

That feels significant. As AI gives board members access to more information, perspectives and analysis, the value shifts towards knowing what to do with it: What deserves another question? What should we be sceptical of? Where is context missing? What evidence sits underneath the conclusion? And when should experience tell us that something simply does not add up?

These are not new responsibilities for board members. But AI may make them more important than ever. AI can give you another perspective. It can challenge an assumption. It can process an extraordinary amount of information. It can help you arrive at a board meeting knowing considerably more than you might otherwise have known. But AI cannot relieve you of accountability for what you do with that information. That remains yours!

There was another result from the Global Forum that I keep coming back to: When we asked how confident participants were that their boards could effectively challenge an AI-influenced management recommendation, many placed their boards somewhere in the middle?

That matters. Management is already using AI. Increasingly, the analysis, forecasts and recommendations arriving in the boardroom will have been shaped by it in some way.

The question for board members is whether they can understand what sits behind those recommendations, challenge the assumptions and biases, assess whether the evidence has been verified, and know when to push further.

This raises practical governance questions too. Should material AI-generated analysis be identified? Should directors be able to trace important conclusions back to the underlying evidence and data? What assurance should exist around AI-influenced board information?

Those questions will become increasingly important. But there is another side of the AI conversation that I worry boards may not be spending enough time on: AI outside the boardroom.

Much of the board conversation around AI has understandably focused on risk: bias, hallucinations, privacy, cybersecurity, intellectual property, explainability and accountability. All deserve serious attention. But while boards are working through how AI should be governed, AI itself is moving rapidly through businesses and markets. It is changing how work gets done, what customers expect, where costs can come out and, increasingly, where competitive advantage sits.

That makes AI a strategy conversation as much as a governance one.

New companies are being built around AI capabilities that many established businesses are still working out how to adopt. Without the same legacy systems, structures and assumptions, some can move considerably faster and challenge established business models.

For boards, that requires a different set of questions:

• Which parts of our competitive advantage become easier for others to replicate?
• Which assumptions in our strategy may no longer hold?
• Where could AI fundamentally change the economics of our business model?
• Who is approaching our industry without the constraints we have learned to accept?
• What becomes possible now that was not economically or technologically possible two years ago?

And perhaps most importantly: Are we spending as much time looking for the opportunity as we are looking at the risk?

Good governance has always required boards to consider both risk and opportunity. The difficulty with AI is the pace at which both are moving.

Perhaps that is the paradox boards need to become comfortable with.

• Inside the boardroom, AI should make us more capable without making us more dependent.
• Outside the boardroom, AI may be changing the business faster than our existing governance conversations recognise.

We need to hold both thoughts at the same time, because the board that spends all its time asking whether AI can be trusted may miss the competitor who has already worked out how to use it. And the board that races towards every AI opportunity without questioning assumptions, consequences and evidence may discover why human judgement still matters.

Perhaps, then, the question is not simply whether boards are ready for AI. It is whether they are asking the right questions about it.

Technology is moving quickly. Our thinking, judgement and preparation need to move faster.