Boards don't need AI expertise to make good AI decisions

There is a widely held belief inside UK boardrooms that goes something like this: before we can approve anything significant on AI, we need to fully understand AI. It sounds responsible. It sounds like good governance. And it can quietly freeze progress while competitors get on with it.

The hesitation is understandable. Nearly three-quarters of boards are perceived to have only moderate or limited AI expertise, according to recent commentary. Boards that carry legal and fiduciary accountability do not want to approve something they cannot assess. That instinct is correct. The misconception is in what "being able to assess it" actually requires.

This article is for Chief Executives, Managing Directors, and Non-Executive Directors who want to govern AI well without becoming AI specialists. It applies equally to the councils and trustee boards of membership bodies, and to partner groups in law and accountancy firms.

Why the expertise myth feels so reasonable

Boards are used to being the most informed people in the room on the decisions they approve. That standard serves companies well. When AI arrived as a boardroom agenda item, the natural response was to apply the same standard: know the subject deeply before signing off.

But AI is a general-purpose capability, not a single investment. Waiting to understand it fully before approving anything is a little like waiting to understand every accounting standard before approving a budget. The board's job is not to do the accounting. It is to ask whether the controls are in place and the risks are proportionate.

The same logic applies here. The question is not "do we understand large language models?" The question is "do we understand what this specific use case does, what data it touches, and who is accountable if something goes wrong?"

What the belief is costing UK SMEs

In owner-managed and board-led SMEs, AI projects frequently stall not because of technical barriers but because decision-makers do not feel equipped to assess the risk. The operational team wants to move. The board is not sure it has enough information to approve. The project sits in a queue.

That queue has a cost. Every month of delay is a month a better-prepared competitor is refining its processes, reducing its cost base, and shortening its response times. Standing still is a more active choice than it appears.

There is also an error-direction problem worth naming. One in four executives report that AI errors have already reached boards or external audiences. That figure suggests the risk of acting without proper oversight is real. But it also reflects what happens when AI is adopted informally, without the board's involvement at all. The answer to that risk is not board paralysis. It is proportionate board oversight applied early.

At gecco, our view is straightforward. AI adoption is 80% people and culture, 20% technology. The board confidence gap is a people problem, not a technical one. Solving it does not require a technology education. It requires a governance conversation.

What the evidence says boards actually need

The evidence points to a much narrower brief than most boards assume.

First, narrow the scope. A board does not need to approve "AI adoption" as an abstract policy. It needs to approve a specific use case, with a named data owner, a defined testing period, and a clear exit criterion. That is a normal business decision. It happens to involve AI.

Second, focus on data handling. The questions that matter are: what data does this touch, where does it go, who can access it, and is that consistent with our obligations under UK GDPR and any sector-specific regulation? Those are questions any competent board can ask and evaluate. They do not require AI expertise. They require the kind of data stewardship discipline boards already apply elsewhere.

Third, establish accountability, not surveillance. The board's role is to ensure someone is accountable for monitoring outputs and reporting back. That person is typically an operational lead, not a board member. The board sets the expectation and receives the report. It does not need to run the process.

Reporting from the manufacturing sector reinforces that the biggest bottleneck in AI adoption is not technology but imagination and governance culture. The same pattern holds across professional services and retail. The technology is often the least complex part of the problem.

A realistic approach for boards ready to move

A practical board briefing on a specific AI use case can be built around three questions: what is the business problem, what is the data footprint, and what does good oversight look like for this case. That briefing should take under an hour. It should produce a decision, not another deferral.

UK boards should also ensure that any AI governance approach aligns with data protection obligations, sector regulation, and internal accountability structures. That is not a reason to delay. It is a reason to scope the first approval carefully and build from there.

For boards that want a more structured grounding before that first conversation, gecco's Training and consultancy programme is designed for exactly this situation. The ESSENTIALS tier runs over three days and covers AI fundamentals, the GRAFT methodology, and practical assistant building. The COMPLETE tier adds autonomous tools and an applied session. The goal is not to make boards into AI specialists. It is to give senior leaders enough competence to ask the right questions, assess risk proportionately, and approve decisions with confidence.

The one thing boards most often get wrong

Boards treat AI literacy as a prerequisite for AI governance. It is not. Governance literacy is the prerequisite. Most boards already have it.

The structured approach matters more than the technology choice. An AI project approved with clear accountability, a defined test period, and a data-handling standard is lower risk than an unapproved project running informally across the organisation. The latter is what tends to produce the errors that reach external audiences.

Start with one business problem. Define what success looks like. Assign a named owner. Set a review date. That is a governed AI pilot. It does not require the board to become AI specialists. It requires the board to behave like a board.

Next step for Chief Executives and NEDs ready to move

If your board is circling the same AI conversation without reaching a decision, the AI Readiness survey is built to surface exactly where the hesitation is coming from and what a proportionate first step looks like for your organisation.

Take the AI Readiness survey. You will get access to 65+ free resources and a custom AI Readiness report. We then offer a free 30-minute showcase to walk through your results.

If you want structured support moving from that conversation to a governed first pilot, gecco's Quick-Start engagement bundles AI Assistants, training, and governance clarity so boards can approve with confidence and operational teams can learn by doing.

Website · LinkedIn · Case Studies · Newsletter