From Awareness to Governance Advantage
Global Directors Council Roundtable – Virtual Co-hosted with Devyani P. Vaishampayan (NED, London) and BoardOutlook.

Insights from Our Co-Host

"Governing AI is challenging for directors, not because they aren't capable, but because of the convergence of several compounding pressures. Technical literacy in areas like machine learning and large language models is not something most directors grew up with or acquired in executive careers. The pace of AI development means boards are perpetually catching up rather than leading. The risk landscape, including algorithmic bias, hallucination, data privacy and intellectual property is unfamiliar territory. And within many exec teams, AI capabilities are still rapidly evolving, leaving boards reliant on external consultants who inevitably carry their own interests. These aren't reasons for inaction. They're reasons to be deliberate."

— Devyani Vaishampayan

Executive Summary

Artificial intelligence is no longer simply a technology investment decision sitting within the remit of the CIO or CTO. It has become a governance imperative, one that is reshaping how boards generate insight, exercise judgement, and fulfil their fiduciary responsibilities.

BoardOutlook recently convened a Global Directors Council roundtable bringing together chairs and non-executive directors from listed companies, government bodies and not-for-profit organisations across Australia, the United Kingdom, Malaysia, and beyond. The discussion revealed a governance landscape in transition: boards that are aware of AI's potential, but still navigating the practical, cultural, and ethical questions that come with embedding it meaningfully into how they work.

Boards that are moving forward with confidence tend to share certain habits. They treat AI literacy as a collective responsibility rather than an individual interest. They separate the question of governing AI in the organisation from the question of governing with AI in the boardroom. They build protocols before they need them, and they approach adoption not as a technology project, but as a governance discipline.

Why AI Is Now a Board-Level Question

For most of the past decade, AI featured on board agendas primarily as a risk or a technology strategy item. Something management reported on, and boards scrutinised from a distance, but that framing is rapidly becoming insufficient.

Directors are now using AI tools to prepare for meetings, interrogate board papers, stress-test strategic assumptions and surface questions they might not otherwise have thought to ask. At the same time, management teams are deploying AI across operations, and boards are being called upon to govern those deployments responsibly. The boardroom is simultaneously a user of AI and a governor of it, and most boards have not yet developed clear protocols for either role.

BoardOutlook's data underscores this moment of inflection. Across approximately 150 skills matrix processes conducted over the past two years, there was close to a nine-point swing of directors moving from "general" to "advanced" self-assessed ratings on technology-related skills, representing a shift largely driven by AI-related urgency. Yet despite this upward movement, not a single skills framework included AI as a standalone director competency. It continues to be treated as a sub-capability within broader technology skill sets. This gap matters. It signals that while directors are individually upskilling, boards have not yet institutionalised AI literacy as a governance capability in its own right.

Two Distinct Challenges, One Governance Response

One of the sharpest conceptual contributions from the roundtable was the importance of separating two distinct challenges that are often conflated.

The first is governing AI. This is the board's responsibility to oversee how the organisation develops, deploys, and manages AI systems, including risk, ethics, and strategy. The second is governing with AI. This focuses on the use of AI tools within the boardroom itself to enhance director judgement, meeting quality, and decision-making. Both demand attention, but they call for different responses.

A useful framework for the latter is to consider AI's role across three phases of the board cycle: before, during, and after the meeting.

• Before: summarising and interrogating board papers, identifying gaps in management reporting, scenario planning, and stress-testing strategic assumptions.
• During: providing real-time analytical support and surfacing devil's advocate perspectives.
• After: supporting decision audits, reviewing meeting quality, and tracking how board thinking has evolved over time.

What unites all three phases is AI's capacity to draw connections across large volumes of information that would otherwise remain invisible, or simply take too long to surface.

Boards that conflate these two challenges tend to get stuck. Governance of organisational AI becomes paralysing because the technology is moving faster than policy can keep pace. The more tractable starting point for many boards is to establish clarity on how AI will be used in the boardroom itself and to build outward momentum from there.

Adoption Is Uneven — and That Carries Its Own Risk

When asked to place their boards on an AI readiness scale, the largest group of roundtable participants — nearly half — described themselves as actively exploring AI in their governance processes. Around a third were aware of or considering AI but not yet acting, while a smaller cohort reported AI as already embedded in how they work as a board. Only a handful indicated no current use at all.

This uneven distribution reflects a broader pattern: AI adoption at the board level tends to be driven by individual directors rather than collective board policy. Some directors are using AI extensively and thoughtfully in their personal preparation; others are not using it at all. The result is a boardroom where some members are arriving better informed, and asking sharper questions than others, without that dynamic being openly acknowledged or governed.

Several participants observed that AI use within the boardroom is currently undisclosed in many organisations. Directors may be using AI to prepare questions or interpret data, while management may be using it to draft papers, yet neither side has openly acknowledged this, and no agreed norms exist. This opacity creates a governance risk in its own right. Without transparency, it becomes difficult to identify where AI is influencing decisions, and whether appropriate human oversight has been maintained.

The practical implication is simple but often avoided: boards need to have the conversation. Agreeing, collectively and explicitly, on how AI will and will not be used is a basic governance hygiene requirement that most boards have not yet addressed.

Cultural Barriers Are as Significant as Technical Ones

Across the discussion, technology literacy featured prominently as a challenge but cultural dynamics emerged as an equally powerful brake on adoption. Boards tend to move at the pace of their least comfortable members, and in many rooms this creates implicit pressure to limit AI use to avoid leaving anyone behind.

The corporate governance function, typically including company secretaries and governance teams, also emerged as a critical and sometimes underutilised lever. These professionals are often cautious about AI adoption, understandably, given their mandate to manage risk and protect the integrity of formal records. However, their involvement in shaping AI protocols for the boardroom is essential. Engaging them as architects of responsible adoption, rather than encountering them as barriers to it, will be important for boards seeking to move forward with confidence.

Directors also noted the value of reframing AI not as a replacement but as an augmentation of what they already do. Early resistance often stems from a sense that AI threatens the judgement and experience that directors bring to the table. Boards that have made the strongest progress tend to have found ways to make AI's value concrete and visible. They connect it to specific decisions, specific questions or specific moments where a tool surfaced something the room had missed.

The Judgement Question

A consistent theme was the need to protect the primacy of human judgement while still unlocking the genuine analytical advantages AI offers. Directors were emphatic: AI is a decision support tool, not a decision maker. The accountability that sits with directors — legal, ethical, and fiduciary — cannot be delegated to an algorithm.

At the same time, several directors shared compelling examples of AI surfacing insights that human review had missed: a pattern in a CEO report that revealed an underappreciated business opportunity; a line of questioning inspired by a modelled analytical persona that unlocked a conversation the board had been circling around for months.

A useful test: if AI had recommended the opposite course of action, would the board have decided differently? If the answer is yes, the AI has replaced judgement rather than informed it. Keeping this question active is one way boards can maintain appropriate oversight of how AI is influencing their deliberations. AI allows directors to consider more things, more widely, and with more data. That is a significant capability. The discipline lies in ensuring it informs rather than substitutes for the judgement the board is there to exercise.

The Risk of Intellectual Complacency

One of the more candid observations from the roundtable concerned the risk that AI makes it too easy to shortcut deep analysis. Preparing for a board meeting through AI summarisation is efficient; it is also qualitatively different from reading carefully and forming independent views. The texture of analysis, the connections made, the doubts raised, the discomfort sat with, can be lost when AI collapses complexity into a convenient briefing.

Directors who have built reputations for rigorous, independent thinking will want to be deliberate about when AI augments that thinking and when it may be subtly eroding it. This is not an argument against AI use; it is an argument for conscious, disciplined use.

A related concern is the tendency of many AI tools to affirm the user's existing position. Without explicit prompting for critical challenge, AI will often find the glass half full. Directors should build the habit of explicitly requesting adversarial perspectives and should be alert to outputs that seem uniformly validating. The tool is, in many respects, very polite. Getting genuine value from it often requires asking it to be otherwise.

Knowing What You're Working With

When a board uses AI to prepare for or support a discussion, directors should understand, at a principles level, where the tool's outputs are coming from — what data it draws on, how it handles sensitive information, and what its known limitations are.

Not all AI tools are equivalent. A purpose-built tool trained on a board's own data operates very differently from a general-purpose assistant drawing on public internet content. Directors don't need to become technical experts, but they do need enough literacy to ask the right questions — particularly before sharing board papers or strategic information with any AI tool.

From Exploration to Governance Discipline

From cyber risk and now AI, each generation of governance challenge has followed a similar arc. Boards begin by treating it as someone else's problem, recognise the strategic exposure, scramble to upskill, and eventually integrate it as a standing discipline. AI is moving through that arc faster than any of its predecessors.

The question for boards is no longer whether to engage with AI. It is how to do so in a way that is deliberate, transparent, and genuinely additive to the quality of governance. That means establishing shared protocols before they are needed. It means treating AI literacy as a board-level capability, not a personal interest. It means engaging the governance function as a partner. And it means being willing to focus AI use on specific, consequential questions, rather than adopting it broadly and hoping the value emerges.

When boards approach AI in this way, it stops being a source of anxiety and becomes what it has the potential to be: a meaningful enhancement to the judgement they are trusted to exercise.

How BoardOutlook Supports AI-Ready Boards

BoardOutlook's governance platform is designed to make AI a trusted, transparent, and contextually grounded resource for boards with appropriate safeguards to house sensitive information.

Our AI capability, OutlookIQ, draws exclusively on data generated through a board's own governance processes: board evaluations, committee assessments, skills matrices and CEO performance reviews. Every insight it surfaces is anchored in the specific history, dynamics, and priorities of that board, not derived from general internet training data. It is built on an institutionalised governance knowledge base developed in partnership with experienced chairs, directors, and governance specialists, which significantly reduces the risk of hallucination and increases the reliability of outputs.

Practically, the platform supports boards across the full meeting cycle. From preparing targeted discussion papers and scenario analyses, to identifying longitudinal trends in board performance, to surfacing questions that data suggests may not be receiving sufficient attention. Every output cites its underlying data sources, so directors can assess the basis for any given insight rather than treating it as an unattributed assertion. OutlookIQ is currently available to all BoardOutlook clients on a complimentary basis as we continue to develop and refine the tool in partnership with our global client community.

This paper was developed by BoardOutlook based on a Global Directors Council roundtable. Participant contributions are reflected thematically and are not individually attributed. BoardOutlook's Global Directors Council brings together chairs and directors from across sectors and geographies to explore governance questions in a confidential peer setting. For information on upcoming sessions or the BoardOutlook platform, please contact the BoardOutlook team at jingqi.wu@boardoutlook.com.