Boards will need “informed engagement” with artificial intelligence if they are to work through the hype surrounding the technology and understand its potential benefits, according to a leading expert.

Sofia Ihsan, AI consulting leader at Forvis Mazars, says despite the urgency to implement AI solutions, particularly genAI, boards need to take a critical approach to manage their expectations of what the technology can deliver.

“Whilst there’s a lot of enthusiasm out there, enthusiasm and excitement aren’t enough for successful adoption,” says Ihsan.

“You need that engagement, and that’s really challenging for boards because they don’t always have the experience, they don’t always have the expertise, and it’s easy to become vendor-led, or to start rushing to deploy things without asking, let alone even answering, the important questions.”

While the initial excitement prompted by genAI may have dimmed, there remains considerable fervour around the potential competitive advantage it can offer.

A report from the OECD in June said AI adoption accelerated in the previous year with implementation “more driven by leaders escaping the pack than laggards catching up”.

Of course, AI is not only genAI. It also encompasses technologies such as machine learning, natural language processing and, of course, agentic AI, a technology that uses other AI tools, such as advanced chatbots able to handle complex customers’ queries and even initiate follow-ups.

Knowing which AI tool to use is part of the challenge for company leaders and a key question for boards too, as they face up to needing to fill key knowledge and skills gaps.

Ihsan argues boards will need training and will need to make strategic recruitment decisions. “You need technologies, you need the ethicists, the cybersecurity people.

“What you can’t do is just give the data to a bunch of data scientists in a darkened room. It needs all these people around the table advising the board.”

Show and tell

One place a board can start in its AI education is establishing full “visibility” of its existing AI through building an “inventory” of technology already in use, while at the same time assessing what is high or low risk. “You cannot govern or engage with something you don’t know anything about,” says Ihsan.

That should be followed by two key questions: are people ready for new technology and do they understand the benefits and the potential risks?

Boards should also ask what the impact might be on the workforce and customers. After all, corporate leaders may have to contend with rolling out AI to a workforce concerned it is about to take their jobs.

When and how?

In many senses, those are AI basics. A more pertinent question for boards is how they balance the pressure to leverage AI now, against taking a considered view of how it can be used and, at the same time, managing the considerable governance implications.

For Ihsan, there are three pillars to answering this question, that she terms purposeful integration, active oversight and dynamic alignment.

Purposeful integration means a close look at which types of AI are actually needed in contrast to “liberally sprinkling” the technology everywhere, in the hope that something works.

“It’s not a magic bullet for everything,” says Ihsan.

Next, boards should consider active oversight: processes and procedures in place to identify and correct when AI, particularly genAI, goes wrong. This is a necessary part of AI implementation because the technology is “dynamic”, constantly updating its own decision-making framework.

Lastly, the dynamic alignment element means boards must keep a close watch on regulatory developments and adjust accordingly. Failure to do so presents a significant risk as rule makers and regulators play catch-up with what the technology can do.

Some jurisdictions—for example, the EU and its AI Act—already have in place significant pieces of rule making that bear close attention. Others, such as the UK, may take a light-touch, “innovation”-led approach, but may yet move on new, more rigid, rules.

Governance in practice

Governance is a challenge. AI legislation overlaps with areas such as frameworks on cyber security, privacy and data protection. This, according to Ihsan, means some companies are starting to build “overall” digital governance processes, bringing the separate and overlapping threads together.

That may be beyond some organisations at the moment, she says.

“For organisations that don’t have that, my advice is: do not reinvent the wheel and set up cottage industries. Harness what you already have in cybersecurity, in data protection, and then just layer on top the additional risks that you’re not already managing.”

There is much to think about, but Ihsan closes with a series of key messages. Prepare company data for AI. Companies “won’t derive the full benefits until the data foundations are there”.

As alluded to earlier, boards should also remember that AI is not just about education and training for a workforce, it’s also about “hearts and minds” as employees confront their fears of replacement.

It’s also essential to remember the governance. “All the research shows that companies that have robust AI governance have better returns on investment; and the companies that best harness AI are also twice as likely to have robust AI governance.”

Perhaps the most urgent message from Ihsan is this: “You can’t avoid this. This is happening, engage with it.”

You can watch the full interview with Sofia Ihsan below