Where to Apply Generative AI to Enhance Board Performance

In 2014 a Hong Kong venture firm, Deep Knowledge Ventures, appointed an algorithm called VITAL to its board. It was largely theater. The software held observer status, had no vote, and most commentators called it a gimmick. The instinct was sound even though the execution was not. A decade on, the idea has matured. Abu Dhabi's International Holding Company, the largest listed company in the UAE, now seats an AI observer called Aiden Insight at its board, feeding directors real-time analysis of risk and strategy.

AI is arriving in the boardroom. The question worth a director's attention is where it genuinely helps, and where it must not go.

The clearest way to see this is as a spectrum. The contribution of GenAI to board work rises from the clerical to the judgmental, and its value rises with it. So does the care it demands.

The clerical floor

At the base sits the work of the record. AI can transcribe meetings, draft minutes and resolutions, and keep consistent documentation across years of decisions. It can compress a board pack that runs to several hundred pages into something a director absorbs before the meeting rather than during it.

This is real value, and it is table stakes. It is also where the first risk appears. A board that accepts an AI summary without interrogating it has quietly handed the framing of its decisions to a machine. The judgment that matters here is knowing what the summary left out.

Where the value rises

Move up the spectrum and AI stops saving time and starts improving thinking.

A board's gravest failures are usually failures of perspective: the unexamined assumption, the option no one raised, the consensus reached too quickly. GenAI is well suited to disrupting exactly this. At the Austrian company Giesswein, researchers placed ChatGPT inside executive meetings and found its real worth lay elsewhere than efficiency. It broke the group's settled patterns, slowed the rush to agreement, and surfaced options the room would not otherwise have considered.

Used this way, AI becomes a tireless source of complementary perspectives. It can argue the other side, apply the stakeholder lens a director is prone to neglect, and generate and pressure-test scenarios the board would never have the hours to build by hand. This is augmentation in its truest form. The board thinks better because the machine has widened and challenged its view.

The apex: the decision itself

At the top of the spectrum is the decision. Here the question shifts from what AI can do to how the work is structured around it.

Most consequential board decisions share a profile. They are irreversible, high in stakes, and hard to assess in advance: strategic direction, a major appointment, a public commitment. For decisions like these, the productive pattern reverses the obvious one. The board forms its judgment first, then has AI stress-test it, challenging the reasoning, red-teaming the assumptions, and surfacing what the room has not seen.

This is Humans + AI at the board table. The machine sharpens the decision. The board makes it. It is the principle behind the most credible board experiments: Aiden Insight, for all its sophistication, sits as a non-voting observer whose role is to inform the directors, never to replace them.

Beyond the meeting

The board does not only meet. It governs in cycles, and the gap between meetings is where the world moves. One of the largest opportunities is continuous intelligence: AI watching for the risk signal, the competitor move, the regulatory shift, and briefing directors before the next session rather than after the damage.

There is a recursive use as well. As management embeds AI across the organization, directors will increasingly use AI to help them oversee it, reading the model risk and governance gaps they could not otherwise scrutinize. The board that governs AI well will be one that has learned to use it.

The line that does not move

Across the whole spectrum, one thing holds constant. The decision, and the accountability for it, stays with the board.

This is not a soft preference. A director's fiduciary duty cannot be delegated to a model, and the dominant failure at this level is the quiet hollowing out of judgment: the rubber stamp on a recommendation no one truly examined. Good governance of board AI is therefore a question of design, woven into how each use is built so that directors stay genuinely inside the reasoning rather than nominally at the end of it.

Handled with that discipline, the value is immense. A board that learns to use AI by degrees, with clerical work automated, thinking widened and challenged, and decisions stress-tested but never surrendered, will deliberate better than any board could before.

That is the opportunity in front of every board today. AI is already arriving. The task is to decide, deliberately, what it does once it is there.