At a recent board meeting, I found myself paying almost as much attention to the process as to the discussion.

There was nothing obviously wrong with the board pack. It had the numbers we needed, the risks that deserved attention, and the strategic choices the board had to weigh. But as we moved through the material, I wondered whether the format itself was beginning to show its age. The company was changing every day, while the board was still viewing the business through a fixed cycle of packs, committees, presentations, questions and minutes.

That is how governance has worked for decades. Directors are part-time stewards. They depend heavily on management for information, meet a handful of times each year, ask questions, challenge assumptions, approve major decisions and hold the company accountable. The model has endured because it serves an important purpose. But AI is beginning to make this rhythm feel more exposed.

Over the past few years, most of the conversation on AI has focused on entry- and mid-level roles. Analysts, coders, call centre agents, designers, researchers and marketers have all been told, in different ways, that AI will change their work. But as generative AI becomes more capable, it is also beginning to reshape roles at the very top, including in the boardroom.

In 2014, Hong Kong-based venture capital firm Deep Knowledge Ventures gave an algorithm a seat on its board. At the time, it was widely seen as an attention-grabbing stunt. More recently, Kazakhstan’s sovereign wealth fund appointed an AI system, SKAI, as a voting board member. Other organisations have begun experimenting with AI-powered board observers. Meanwhile, one survey of global CEOs found that 94% believed AI could provide better counsel than at least one of their current board members.

These developments have fuelled a provocative question: with AI in play, will boards become obsolete?

The question is not absurd. Boards have always faced an information disadvantage. Directors are not inside the company every day. They receive information that is selected, organised and interpreted by management. They are expected to make sense of increasingly complex organisations within the limited time available to them.

AI appears unusually well suited to close part of that gap. It can read at speed, compare widely, simulate scenarios, identify patterns and ask difficult questions without worrying about hierarchy, relationships or boardroom politics. If the board’s role were only to process information and produce recommendations, the threat would feel more immediate.

But boards are not just information-processing bodies. They provide accountability to shareholders, exercise stewardship over long-term value and make judgements that involve ethics, trust and legitimacy. These roles cannot be outsourced to machines. AI can illuminate options, but it cannot own the responsibility of choosing between them.

This is partly because the inner workings of large language models are still difficult to interpret or audit. As a piece published in Harvard Business Review points out:

What exactly [LLMs] internalize—and whether those patterns truly align with human intent—remains opaque. For governance, this raises both promise and risk: AI boards might gradually pick up elements of judgment and nuance, but in ways that are difficult to trace or control.

That opacity matters in a boardroom. Governance cannot depend only on answers that are fluent, fast and plausible. It also needs traceability, challenge, accountability and the ability to explain why a decision was made.

So the view that machines will make human directors obsolete is extreme. But it would be a mistake to dismiss the provocation too quickly.

This week, let’s unpack how boards can govern effectively in an AI-enabled world. How can AI support directors in fulfilling their roles? What new responsibilities does AI place on boards? And what should board members do to keep adding value in the room?

The Case for AI Literacy

Every era redefines the capabilities expected of directors. Following the collapse of Enron, boards came under pressure to strengthen their financial expertise and oversight. Today, it is difficult to imagine a director arguing that understanding financial statements falls outside their purview.

AI literacy is moving in a similar direction.

This does not mean every director needs to become a machine learning expert. But it does mean understanding enough about AI to ask better questions about strategy, risk, talent, competition, ethics and long-term value. A director does not need to know how every model is built. But they do need to understand how AI could change the company’s cost structure, customer proposition, workforce model, sources of advantage and exposure to risk.

That distinction matters. The board’s role is not to manage AI implementation. That remains management’s job. But the board does need to understand whether AI is being treated as a small productivity tool, a major operating lever, a product capability or a force that could reshape the business model itself.

The need for AI-fluent boards is becoming more urgent. A 2025 MIT study found that businesses with AI-savvy boards surpassed their peers by 10.9 percentage points in return on equity, while those without lagged 3.8 percentage points below the industry average. At the same time, a global Deloitte survey found that 66% of directors reported their boards had “limited to no knowledge or experience” with AI, while nearly 31% said AI was not even on the board agenda.

That gap should worry boards. It suggests that AI is entering the business faster than it is entering the governance conversation.

This is where many boards will need to be honest with themselves. Reading a few articles is helpful, but it is not the same as fluency. Listening to one external expert at an annual offsite may start the conversation, but it will not change how the board works. Adding one technology director may help, but it cannot serve as an excuse for everyone else to remain safely uninvolved.

The chair has a particularly important role here. As an HBR article on pioneering boards using AI notes:

Sustained personal commitment to AI on the part of the chair is critical to maintaining momentum. If board members see the chair learning and applying the new technology and making it a part of the board’s ongoing conversations, they will pay attention.

That is an important point. Boards take their cues not only from what is formally on the agenda, but from what the chair treats as important. If AI is handled as an occasional technology update, the board will engage with it episodically. If the chair keeps bringing AI into discussions on strategy, risk, talent and competition, the conversation starts to shift.

AI Raises the Standard for Contribution

AI literacy is important not only because boards need to understand a new technology. It matters because AI will raise the standard for what counts as meaningful board contribution.

Most boards already know the difference between a director who adds value and one who merely attends. The valuable director has read the material, but is not trapped inside it. She notices the assumption that management has made but not stated. He connects the discussion to another industry or cycle without turning every conversation into a story from his own past. They ask questions that improve the quality of the decision, not questions designed to prove they have read the pack.

AI will not replace that kind of contribution. But it will make weaker contribution harder to hide. If a tool can summarise the board pack, compare competitors, flag inconsistencies and generate a list of sensible questions, then the human director has to bring something more than basic preparation.

The premium shifts to judgement, context, courage and the ability to sense what is not being said.

This is why the AI discussion is not only about technology. It is about board effectiveness. It asks each director a more personal question: am I adding perspective that would be hard to get elsewhere, or am I mainly reacting to material prepared by others?

This is especially relevant because boards are already under pressure to improve their own effectiveness. PwC’s 2025 Annual Corporate Directors Survey found that 55% of directors believed at least one member of their board should be replaced. The top reason cited was lack of meaningful contribution to board discussions. That finding has implications well beyond AI, but AI will sharpen it. When information becomes easier to access and routine analysis becomes easier to generate, the board member’s real contribution becomes more visible.

Past credentials will still matter. Former CEOs, finance experts, sector veterans, lawyers and regulators will continue to bring valuable experience. But experience has to stay alive. In an AI-shaped world, the most valuable board members will be those who combine experience with curiosity, judgement with learning, and confidence with enough humility to keep updating what they know.

Why AI Cannot Sit in a Technology Corner

For years, technology oversight could often be delegated to CIOs, special committees or occasional expert briefings. Gen AI changes that dynamic because its impact does not stay inside technology. It touches strategy, customer experience, product development, workforce planning, intellectual property, compliance, cyber risk and culture. It can change not only how efficiently tasks are performed, but also how decisions are made inside the company.

That is why AI has to become a board-level issue.

The risks are also becoming more complex. Bias, fabrication, IP leakage, cyberattacks, legal exposure, energy intensity and over-reliance on automated outputs are all real concerns. But the deeper risk is more subtle: companies may begin embedding judgement into systems without being clear about where human accountability sits.

That is a board question. When an AI-supported process produces a poor decision, who is responsible? When productivity gains come at the cost of trust or capability, who decides whether the trade-off is acceptable? When vendors become deeply embedded in critical workflows, is the company building advantage or quietly giving it away? When employees see AI being rolled out aggressively, do they feel augmented, monitored or replaceable?

Boards do not need to answer all these questions themselves. But they do need to ensure that management is asking them seriously.

A Practical Guide for Boards

The practical response is not for boards to become more intrusive. That would be the wrong lesson. The role of the board is not to become a shadow management team. The role is to improve the quality of governance.

Here are seven areas where boards can start.

1. Move from AI updates to AI posture.

Many boards are still at the stage where management presents a periodic AI update: pilots underway, use cases being tested, productivity opportunities identified and risks being monitored. That is a reasonable starting point, but it is not enough for long.

Boards need to ask a more fundamental question: what is AI becoming in this business?

Is it mainly a tool for optimisation? Is it being used to expand into new offerings or customer experiences? Is it becoming part of the product itself? Could it reshape the company’s operating model or business model? Is the company trying to lead, follow carefully or learn selectively before committing?

The governance model should differ depending on the answer. A company using AI to improve call-centre productivity does not need the same board engagement as a company building AI into the core of its customer proposition. A business experimenting in small pockets should not be overseen in the same way as one making enterprise-wide AI investments.

Without clarity on posture, AI governance becomes vague. Everything sounds important, but the board struggles to know where to spend its time.

2. Build AI fluency across the board.

The challenge posed by AI cannot be delegated to a single expert or committee. Every director must build a working understanding of the technology, its impact on competitive dynamics and how it can change the business for better or worse.

This may involve formal education, hands-on experimentation, external briefings and direct engagement with the people working on AI inside the company. Directors should not only learn about AI in abstract terms. They should understand where it is being used in their own business, what value it is creating, where it is struggling and what risks are emerging.

The board should also make learning practical. A general session on AI may create awareness. But directors are more likely to build fluency when they experiment with approved tools, apply them to board preparation, compare outputs, discuss what worked and understand the limits.

3. Use AI to prepare better, not think less.

Once the board is comfortable with AI, directors can start using it to enhance preparation and deliberation. AI can help surface relevant insights in the board pack, examine external context, benchmark competitors, identify market shifts and stress-test proposals.

The real value, however, is not in producing more analysis. It is in improving the quality of questions.

Before a meeting, a director could use approved AI tools to ask: What assumptions is management making? What external developments could make this plan wrong? What would have to be true for this proposal to work? Which customer signals or competitor moves deserve more attention? What risks might be underweighted because the current performance looks strong?

This is where AI can become a good thought partner. It can help directors widen their field of view before they enter the room. But it should not become a substitute for their own judgement. A director who uses AI well will come better prepared, not less responsible.

4. Move board packs from reporting to insight.

Many board meetings still spend too much time reviewing information and not enough time discussing what the information means. AI creates an opportunity to redesign that dynamic.

Routine summaries can be automated. Variance analysis can become faster. External benchmarking can be made more accessible. But the goal should not be longer packs, more dashboards or more impressive-looking charts. Most boards already have enough material.

A better board pack should make it clearer what has changed, what management believes, which assumptions matter, where there is genuine uncertainty and what requires judgement from the board. It should surface weak signals before they become formal agenda items: a shift in customer complaints, a change in employee sentiment, an emerging competitor pattern, a regulatory development, an unusual movement in service quality or retention metrics.

Board packs should not only tell directors what happened. They should help directors understand what deserves attention.

5. Decide what belongs where.

As AI becomes more central to the business, boards will need to be more explicit about what belongs with the full board, what belongs in committees and what should remain with management.

A recent McKinsey paper offers some guidance:

Explicitly define which topics should be reviewed and fully discussed in full-board sessions (for example, material investments to scale enterprise-wide AI), which belong in committees (for example, risk frameworks and material vendor reviews), and which do not require significant board discussion (such as regular operational decisions).

AI can otherwise become everyone’s topic and no one’s responsibility. Material investments, business model implications and major risk appetite choices should come to the full board. Risk frameworks, vendor reviews, cyber exposure, compliance and auditability may sit with committees. Operational decisions should not clog the board agenda unless they carry strategic or risk implications.

The board’s task is to ensure clarity of ownership without taking over management’s role.

6. Engage more broadly with the people doing the work.

Boards should hear from the CEO, CFO and CIO on AI. But that is not enough. AI is being embedded in products, customer journeys, operations, risk systems, marketing, HR and finance. Directors need exposure to the people actually doing this work.

This could include product leaders, data teams, risk leaders, customer-facing teams, HR, legal and business unit heads. It could also include rising leaders who are closer to adoption challenges than the senior team may be. Without these conversations, AI remains too abstract. The board hears the ambition and the risk summary, but not the friction of implementation.

Boards should ask: Which pilots are scaling? Which ones are being stopped? Where is value being created? Where are controls still weak? How are employees responding? What capabilities are we building internally, and where are we becoming dependent on vendors? Which risks are showing up in the field before they appear in formal dashboards?

These conversations help boards connect strategy with operating reality.

7. Set protocols for board-level AI use.

Using AI in the boardroom requires rules of its own. Directors cannot paste confidential board materials into public tools because they want a quick summary before a flight. The board should work with general counsel and legal advisors to establish clear protocols around approved platforms, data security, confidentiality, record retention, legal privilege and human review.

Sensitive board materials should remain within secure, company-approved AI environments. Guardrails should be developed to mitigate hallucinations, bias and over-reliance. Directors should know which tools they can use, what information they can upload, what outputs need to be verified and how AI-assisted preparation should be handled.

This is not bureaucracy for its own sake. It is basic governance hygiene. AI can support board deliberation, but human reasoning must remain paramount in final decision-making.

The Human Work Becomes More Important

There is a temptation to make the AI conversation entirely about tools, models, vendors, pilots and productivity. Those things matter. But the deeper board conversation is about judgement.

Where should the company move faster, and where should it slow down? Which decisions should remain clearly human? What level of error is acceptable when AI is used at scale? How should employees be reskilled, not just redeployed? What should the company refuse to automate even if it can? How will customers know when they are interacting with AI? How will the company protect trust while pursuing efficiency?

These are not technical questions alone. They are questions of stewardship.

AI can help management move faster. It can help boards see more. But speed and visibility do not automatically create wisdom. In some cases, they can create false confidence. A fluent answer can still be wrong. A precise forecast can still rest on a fragile assumption. A productivity gain can still damage culture. A clever AI use case can still distract from the harder question of whether the company is building real advantage.

That is why human judgement does not become less important in the AI era. It becomes more visible.

What Should Boards Do Now?

The best starting point may not be a grand AI transformation of the board. It may be a more practical conversation at the next board meeting.

What is AI becoming in our business: a productivity tool, an operating lever, a product capability or a reinvention opportunity? What part of our board process has not changed in years? What information reaches us too late? Which assumptions do we accept too easily? What should no longer take up board time because AI can help do it better? Where does AI create new risks that our current governance model may not catch? And where does human judgement matter even more because AI is now entering the decision process?

Boards do not need to fear AI. But they do need to take seriously what AI is revealing about their own relevance.

AI will not replace the board’s responsibility. But it will challenge the board’s usefulness. It will reward boards that learn faster, ask sharper questions and redesign how they work. It will expose boards that treat AI as a passing technology update while the business around them changes more fundamentally.

The opportunity is not to make governance more automated. It is to make governance more intelligent.

That may be the real AI reckoning in the boardroom.