So… what are we doing with AI? Innovating in an age of caution

It is a situation many CEOs are familiar with. Seated at that most intimidating of tables, fixed by the steely glare of their board members, faced with the question: “So…what are we doing with AI?”

Recommended Video

Artificial intelligence has gripped the business community’s attention like few technologies before, but the payoff is far from guaranteed. More than half (56%) of CEOs participating in PwC’s latest Global CEO Survey said their AI investments have yet to produce any meaningful financial benefits, with only 12% reporting both cost efficiencies and revenue gains.

When new technologies promise solutions to persistent problems, leaders are faced with a difficult choice. They must find ways to explore these options without sacrificing short-term performance.

This tradeoff is harder than ever. PwC’s study also showed that only 30% of CEOs say they are confident about revenue growth in 2026—the lowest this number has been in five years. In a time of geopolitical instability, committing to major investments can feel extremely risky.

Nevertheless, “if you don’t innovate, you could die,” says Joe Petyan, U.K. CEO of WPP-owned advertising agency VML. The challenge for leaders today is not whether to innovate—but how to experiment without destabilizing the very business they are trying to transform.

Inside the boardroom

Disagreements with the board are a routine part of C-suite life, but the transformative potential of AI—and the speed at which successful companies are moving from experimentation to application—has intensified the pressure significantly.

“What I’m hearing a lot more from industry peers is, ‘The board has gone from being interested to being demanding when it comes to AI,’” says Ronan Harris, president of EMEA at Snap Inc. “‘I need to show up with demonstrable results, but I’m also being told I can’t sacrifice my targets.’”

Despite this urgency, most companies are still early in their journey. Only 10% of organizations report high levels of AI maturity, according to KPMG’s 2026 Global Tech Report. Boards are looking for progress even as firms are learning to deploy the technology effectively. But, says Harris, many leaders “are not clear where to find the resources to experiment.”

Risk appetite within business isn’t static—it evolves. Shail Deep, chief operating officer for EMEA and APAC at global data and technology company Experian, highlights three distinct eras of risk attitude in just the past 20 years. After the 2008 financial crisis, she says, it was about avoiding all risk. Then conversation shifted to assessing which risks were worth taking. Now it is more complicated still. “It has become a much more dynamic, strategic conversation,” she says. “If you’re too fast and reckless, you erode trust. If you’re too slow, you get left behind. It’s about how fast an organization can move strategically.”

Structural, not symbolic

This strategy must start in the boardroom. Unless everyone begins an innovation project on the same page, undue pressure to make tradeoffs is inevitable.

Harris points to past examples of this. “When the internet came along, the people who got innovation wrong—very badly wrong—were the folks who said, ‘Digital computers, internet…sounds like an IT thing. Give it to the IT guy.’”

The right approach, he says, is to commit to making innovation projects a focus of board meetings, with a specific set of KPIs and a sense of how the organization is benchmarked against the external market.

Ryan Dullea, chief growth officer at multinational consumer goods company Reckitt, agrees. Reckitt’s approach is to develop the overarching strategy with the board at the start of the year and to agree, explicitly, what sort of innovation projects ladder up to that strategy’s goals. “Then, when we bring ideas and investments through to the board, they line up with the strategic intent we’ve agreed on.”

There are two steps to ensuring any innovation will meet with board approval. The first is to quantify whether the potential reward is worth the risk. The second is to build in checks and balances from the outset. Many companies now treat innovation investments more like venture capital portfolios— funding ideas in phases, rather than committing everything upfront. “You need to have a number of stage gates throughout the process,” says Dullea.

The power of sandboxes

Once board approval is secured, the real work begins. One increasingly common approach is to create protected environments for experimentation, known as sandboxes.

This involves creating mini-startups within the organization, with dedicated budgets and staff. These teams are empowered to experiment. This helps avoid the common problem of getting stuck in the pilot phase, a challenge facing many companies, a recent report by Deloitte found. According to its 2026 State of AI in the Enterprise, only 25% of organizations have moved 40% or more of their AI experiments into production to date. Sandboxing allows firms t