How boards can enable AI profitability

Ask the right questions

Board members don’t need to be AI experts to discern which potential AI cases are worthy of effort and investment. Carlsson advised board members to ask management two simple questions:

• How is this going to deliver value for us?
• How are you going to execute on that?

If the answers to those questions aren’t both clear and satisfactory, it’s time to ask for more details. Board members have substantial experience in asking the probing questions that will surface necessary issues, and those same skills will be essential in AI oversight — particularly in separating the wheat from the chaff.

Carlsson said boards also have the responsibility and power to ensure that management’s attention, resources and focus are channeled toward meritorious use cases. At the same time, boards should inquire about and advocate for the individuals within the organization who have the technical ability and organizational understanding to discover and implement AI use cases that are capable of delivering tremendous value.

These innovators understand the importance of aligning AI use cases to business needs, as well as the often-missing quality requirements at the outset of AI projects. When boards identify such visionaries, they need to advocate for them. “These are individuals who are struggling for resources right now,” Carlsson said. “They’re making tempered, realistic predictions about what they’re going to deliver. … [If you support them], you’ll be their best friend, because they will be able to so much more effectively change the organization.”

The backing of the board also can create additional impetus for these individuals to show tangible business results for their projects.

To build or to buy

Once organizations have established a business case for AI, they are often challenged to decide whether to build or buy a solution. Both choices come with challenges, Carlsson said.

Proven off-the-shelf solutions can produce quick, low-risk returns for common use cases while establishing a strong data foundation for more advanced AI use cases. But these tools that can be purchased might not be suited for the unique purposes of an organization’s differentiated business needs.

“The solutions that you can buy are not going to [produce] disruptive use cases for your organization,” Carlsson said. “They’re going to be for the low-hanging, most generic use cases. Customer service automation, for example.”

For more differentiated use cases, Carlsson said an organization’s processes, systems and data are often distinct enough to warrant building AI solutions rather than buying them when the appropriate resources are available. The problem with the “build” option is that it may lead to a jumbled ecosystem of different nascent technologies that won’t integrate well with one another over time.

Without governance, various departments throughout an organization might build their own solutions, and this results in what Carlsson calls “the jungle” of technological complexity. Boards have an opportunity to make the jungle less fearsome because they have a comprehensive view of the entire organization and can work with management to develop an understanding of where connections can be made.

• Engage in skip-level meetings. Executives in the C-suite often are too far removed from the mechanics of AI implementation to understand where the most value can be generated. Through skip-level meetings, board members delve into the organization and interact directly with the individuals who can shine the brightest light on the organization’s most promising AI use cases. They might be chief data officers, newly elevated chief AI officers, or even departmental innovation leads. Whatever the case, these should be the people who most understand the biggest opportunities for AI and the investment that’s needed. These meetings have the added benefit of helping board members upskill their understanding of AI.
• Build knowledge about real use cases. Board members can speak with people from other organizations about the mechanics of successful projects and the pitfalls and challenges of unsuccessful efforts. Boards don’t need a deep technical understanding, but they can provide better oversight if they understand how the technology solves business problems. Third-party assessments of potential company-specific AI opportunities also can help boards — and management — find a path forward with the best technology fit for a given AI use case.
• Break down silos. Everyone wins when use cases that work in one department are shared throughout the organization. Board members can make a difference by ensuring that management is fostering cross-functional communication toward this end, and a center of excellence often is an excellent vehicle for this.

When interest in GenAI was first accelerating, Smith reminded board members at the NACD Summit that directors have plenty of experience at providing oversight that helps organizations successfully adopt new technology. As boards have embraced that role, Smith said board members now are eager to see the businesses they represent reap the rewards of AI.