The AI governance question boards are getting wrong

Ask the C-Suite what the company’s AI strategy is, and boards will get a confident, polished answer. What they rarely get is the full truth: a good answer to the wrong question tells them little about how AI is actually being governed.

Julie Averill, former CIO of lululemon turned board director, tells UNLEASH that boards don't need to reinvent the wheel with AI. They simply need to ask the same question they would ask about any boardroom topic: “What are your business strategies?”

The next question is: “How can AI help you achieve that better, faster, cheaper?”

By focusing on that question instead of the business one underneath it, Averill says, organizations are just “whitewashing”: signalling that they’re on top of a “technology they don’t fully understand,” rather than actually governing it.

Dan French, CEO at tech consultancy Consider Solutions, agrees with Averill. In his view, the big mistake that boards make is “forgetting to start with the business problem you are trying to solve,” before figuring out where technology plays a role.

Academic Director of INSEAD’s Corporate Governance Center, Annet Aris, adds that there’s a risk of boards focusing too much on “fancy AI tools,” which means missing the bigger picture.

“Implementing the AI tools themselves are not the big challenge – the big challenge is how do you transform the organizations to work effectively with AI,” she tells UNLEASH.

This means asking questions like: “Who is taking which decisions? How are we going to redefine the workforce? What will be the new operating model?”

New technology, but the same governance challenge

There’s no doubt that AI presents fresh challenges for boards.

INSEAD’s Aris notes that the speed of development is challenging for boards to keep on top of – there’s a fear of “obsolescence” if organizations don’t capture the opportunities from AI.

This is pushing many organizations to see AI as a silver bullet – they aren’t thinking about where to use AI effectively, they are simply applying it everywhere.

French thinks that instinct misunderstands what AI actually is. Conventional business software runs on deterministic rules – fixed logic that produces the same result every time. By contrast, AI is probabilistic: it works on predictions, not certainties. “Probabilities are useful,” he says, but that also means AI isn’t the right tool for every part of the business.

“Where you should apply probabilistic thinking is something that people don’t even stop to think about,” notes French.

By his estimate, only 30-40% of business problems genuinely benefit from a probabilistic, AI-based approach – the rest are better served by deterministic technology. Getting that split wrong, he argues, is a governance failure as much as a technical one.

To move the needle on governance, French, therefore, calls on boards and organizations to stop seeing AI as an “activity:” how much employees use AI is “no measure of success”.

“Smart companies say we’re all in on AI where it makes sense,” he says.

Averill agrees. She tells UNLEASH that “boards are feeling the pressure on AI without fully understanding the outcomes they’re searching for.”

Focusing on outcomes, rather than activity, is Averill’s advice to boards on governing AI. She calls on them to “start with oversight, not operations.”

Aris notes that it can be tempting for boards to take on the day-to-day management of AI, but that’s the job of the company’s C-Suite.

Boards “have to shift their framing from ‘do we have AI’ to ‘where is AI making consequential decisions in this company’,” continues Averill.Boards “have to shift their framing from ‘do we have AI’ to ‘where is AI making consequential decisions in this company’,” continues Averill.

Their job with AI is the same with all their other governance priorities; they need to be “asking the right questions that leads management to do its job to create the guardrails that the company needs,” Averill adds.

Where does HR fit into AI governance?

As AI disrupts businesses, “all leaders need to be part of the discussion about how we make our business better,” states French. This includes CHROs and the HR function.

Aris notes that CHROs are no longer just leading the people function, they now have responsibility for bigger picture challenges like transformation, change management and leadership development. They are required to talk to the boards directly on these topics.

French argues that HR’s responsibility when it comes to AI is reminding “people what the rules of the game are”; this requires training and policies around only using company-approved tools when using AI at work.

It can be damaging to the company if employees embrace “random use of public LLMs without thinking carefully about what information you’re putting in. Nothing is free; if you can’t find the product, it’s probably you” and your data input.

He advises HR to remind employees to “not subjugate your critical thinking to any technology.” HR needs to go “all in on harnessing the creative genius” of its people.

French concludes: “If all we do is rely on AI for creative genius, you’ll get the same creative genius as everyone else – and go out of business very quickly.”French concludes: “If all we do is rely on AI for creative genius, you’ll get the same creative genius as everyone else – and go out of business very quickly.”

Not everyone agrees on how far HR’s remit should stretch. While Aris sees the CHRO role potentially expanding to include AI agents as well as human talent, Averill draws a harder line.

She acknowledges that HR has responsibility for “making sure that the leaders and the culture is set up to accept, adopt and embrace AI,” but when it comes to AI governance itself, “I personally don’t think that’s the role of CHROs.”

Rather than parking responsibility at HR’s door, Averill believes that accountability should sit in each business function.

“If I’m the CFO, I should know every person and every agent that’s making decisions” in the function. “I’m ultimately responsible for that.”

“If AI makes a bad decision for supply chain (for example), and you start shipping products to the wrong country. Whose responsibility is that? It’s the supply chain leader.”

“HR has a broad, not very deep, view of the business – they have a basic understanding of every function, but they cannot understand it as deeply as those closest to the work,” concludes Averill.