6 questions for boards: AI and human capital

Support for AI progress goes far beyond risk management

In AI risk management, the biggest risk to many organizations might be failing to take advantage of AI’s transformative capabilities.

This risk should be top of mind for boards as they perform their governance and oversight duties over AI. Their savviest competitors are almost certainly adopting AI, driving efficiency and providing insights that can make their organizations more profitable.

Many boards initially saw their AI-related responsibility as urging management to develop guardrails to protect organizations from the risks related to bias, data security and transparency that are associated with the technology. But it’s long past time to go beyond a protective stance related to AI and take a proactive approach toward the technology.

“The board can and should encourage management to proceed with AI initiatives and be comfortable with being uncomfortable once they have strong risk management and controls in place related to AI,” said Grant Thornton Growth Advisory Services Managing Director Joe Ranzau.

AI, in particular, has the potential to significantly improve the effectiveness and efficiency of human capital at organizations across the industry spectrum. Employees can use AI to multiply their impact many times over while also eliminating repetitive, low-value tasks that don’t require human judgment. However, optimal use of AI requires a commitment to user training, and boards have a responsibility to monitor AI’s effects on employee engagement.

Ranzau, who moderated a recent NACD webcast, “Empowering private company boards: AI and workforce,” provides a deeper dive on AI’s impact on human capital in this Q&A, with an emphasis on the board’s role as an AI champion.

How can boards make sure management provides employees with enough room to experiment with and pilot AI while maintaining appropriate guardrails?

In some cases, management isn’t especially supportive of AI use, and boards need to push leadership forward.

Although 60% of finance leaders in Grant Thornton’s CFO survey for the fourth quarter of 2024 said their organizations are using generative AI, that means 40% weren’t using it yet, even though most of them are exploring potential use cases. Especially when management is reluctant to implement AI, boards need to set expectations for moving forward with AI, establishing a supportive environment with proper resources for AI initiatives that align with the organization’s strategy and ethical standards.

Management must be ready for AI use to advance throughout the business environment — and it’s advancing quickly. It’s the board’s role to make sure that management provides employees with AI tools and training needed to succeed with the right controls in place to remain within the guardrails.

These elements sometimes aren’t in place. In our most recent CFO survey, two-thirds of respondents who are using generative AI said they have clearly defined acceptable use policies for the technology. That leaves fully one-third of these organizations whose people can’t be certain when they can use generative AI — and when it’s off-limits.

Boards that allow management to be complacent on this issue should be forewarned: If the companies don’t provide guidance, there’s a good chance their employees are using AI anyway — and they might even be paying for their own AI tools. Boards must make sure that management is providing employees with clear direction on when it’s OK to use AI and when it’s not. Proper oversight by the board is required to make sure organizations provide employees with an organizational framework for AI — and abundant opportunities within that framework for employees to experiment with AI.

How can boards verify that management takes the right steps to upskill the workforce related to AI?

There is certainly work to be done here. Our latest CFO survey shows that 57% of finance leaders do not plan to increase their spending on training, learning and development in the coming year, while 65% of finance leaders expect to increase their spending on IT and digital transformation.

“There’s not one answer to this because organizations are in different phases of maturity related to AI,” Ranzau said. “The 20% that are early adopters have been using generative AI for a while, and the 40% that are the fast followers also are using it. But that last 40% is still in the adoption phase, where they’re just starting to figure out how to use the technology.”

Although it’s difficult to generalize because of the wide range of maturity levels, it’s important for companies throughout the AI maturity curve to make a concerted effort to upskill their people as they implement generative AI. Training should be thorough and include everyone who will need to use the technology. And it should include detailed instructions for employees to incorporate AI into the use cases that already have been identified as well as a framework for experimenting with AI.

Boards need to make sure that management is allocating resources to upskilling and is defining the skills that will be needed for success with AI. Management should be reporting back to the board on the effectiveness of upskilling efforts, as well as on the workforce’s preparedness to make the most of this important new technology.

How should boards approach oversight of culture related to AI governance? Employees in some situations are concerned that AI is going to take their jobs, and it seems that it would be important for boards to make sure management is transparent with employees with frequent communication on why AI is being adopted and how it will affect their roles.

AI is having a real impact on roles. In the most extreme examples, the need for accounts payable employees and frontline call center representatives has been substantially reduced as AI emerges. It’s important for boards to work with management to set the right tone at the top and be transparent about the way jobs are being affected by AI, even though widespread job displacement is not yet happening.

There are some standard tools that boards can encourage management to use to monitor the culture of an organization. Boards can make sure that management is constantly measuring the pulse of an organization through employee surveys, town halls and focus groups. Boards also can enact skip-level meetings, where they drop down beneath the levels of the organization they might ordinarily meet with, to see how AI is affecting an organization’s culture.

However, employee concerns about AI aren’t occurring in a vacuum, and they are part of a broader awareness that management and boards need to have about culture in the current work environment. Grant Thornton’s State of Work in America survey for 2024 showed that 51% of the U.S. workers who responded said they had suffered burnout in the previous 12 months — a 15-percentage-point increase over those who reported burnout in 2023.

Burnout is high and engagement has surpassed historical lows, and it’s partly because of the removal of boundaries between work and home over the past several years — dating back even before the pandemic. Even people who work all day in an office have laptops and smartphones that demand their attention outside of working hours, and the separation between work and home continues to erode.

In addition, factors outside of work are causing burnout. Employees are experiencing anxiety over economic disruption, political disruption, global conflicts, and climate change. The stress on individuals is incredibly high, and now AI’s emergence is requiring employees to learn how to do their jobs differently, too. Boards might suggest to management that AI:

• Has the potential to alleviate some employee burnout by creating efficiency in day-to-day tasks that can make work less overwhelming.
• Can increase employee engagement as they find creative ways to interact with the technology and experience its benefits.

Oversight of these cultural issues is part of the board’s risk management duties, as a healthy workforce is critical to an organization’s success regardless of the extent of AI adoption. It’s important for boards to closely monitor employee engagement statistics and urge management to take the necessary steps to keep morale high.

How can boards make sure that management responds appropriately to employee resistance to AI?

One thing boards can do is make sure management and labor collaborate on a continuing basis to evolve the organization in a healthy manner. For example, the recent labor negotiations with longshoremen are typical of an issue that some organizations will face in the coming years.

There’s a lot that’s about to change in the internal audit and compliance space. Some percentage of internal audit’s work will continue to be compliance, but another part is going to be devoted to helping improve the performance of the organization. These duties will be related to understanding the technologies that exist and making sure the organization is aware of the potential risks or liability.

The board’s role in this is understanding that it’s not enough to make sure the CIO, or chief technology officer, is upskilled with AI’s capabilities in mind. The audit committee needs to make sure the chief audit executive is upskilled and has people on the internal audit team who understand what they need to be looking at related to AI and how they should be thinking about it.

They need to make sure internal audit is prepared to support the organization’s use of AI in an effective, ethical and legal manner that brings value to the organization. For a lot of management and internal audit teams, that’s a substantial shift from where they have focused in the past.

How can boards make sure return on investment is being measured appropriately for human capital initiatives?

Some board members have said that measurement of AI isn’t necessarily different from a board perspective from how they would measure the return or success of any other transformation or technology initiative.

Typically, that means boards should require management to report on key performance indicators that might include cost savings, productivity gains and even revenue growth. And these metrics should be reviewed regularly and presented to the board. But that’s not always easy with AI.  

From management’s perspective, the challenge with AI is that when AI is experimental, it’s difficult to identify where there are opportunities to measure AI’s impacts. So, for example, in the case of a large language model that employees use to assist brainstorming, it’s probably fairly difficult to measure the return on that brainstorming facilitation.

On the other hand, if you’re using AI, for example, to monitor for fraud in the HR space, whether it’s labor-related, overtime abuse, or shadow employees, you can measure those results specifically compared with how your old compliance programs delivered.