Since ChatGPT exploded into the zeitgeist in late 2022, business buzz around it has been rife with sometimes over-the-top declarations about the scope of the role that generative AI will play, including dire warnings for finance leaders about the AI skills they need to obtain to keep their jobs. 

But exactly what AI skills do finance executives need, and how can they develop them? CFOs and other finance leaders may be happy to know those skills don’t mean learning to code your own AI models. They do, however, mean understanding how AI works and what it’s capable of doing, said KPMG’s US Consulting AI Leader Joseph Parente. 

“You don’t have to be a software engineer or even know how to write Python code. But you do need to understand machine learning concepts,” Parente said in an interview. “How does machine learning and natural language processing actually work? What is possible, what is not possible?” 

Octavio “OJ” Laos, director of AI for California-based accounting firm Armanino, has sometimes been astounded by the misconceptions finance leaders hold, and so recommended before anything else that they try out some of the publicly available and often free generative AI models just so they can wrap their minds around how it works in real life.

“What they need more is a simple actual interaction with it, maybe not so much knowledge as experience. It’s hard to conceptualize what is at stake until you use some of these tools,” he said in an interview. 

CFOs must link AI to goals  

For the CFO specifically, they must learn how to relate AI to the broad strategic goals of their organizations, and a big part of that is finding good use cases that fit their individual circumstances, according to W. Michael Hsu, CEO and founder of California's DeepSky, a CFO consulting and outsourced accounting company.

“For CFOs, AI becomes a sandbox for strategic thinking. The CFO’s role is no longer just about managing financial data; it’s about collaborating with AI to explore new models, test ideas rapidly, and make better decisions faster,” he said. 

To this end, a CFO might need to learn how to leverage AI to interpret data and detect anomalies, or to do scenario modeling for testing strategies and ideas, Hsu said.  

KPMG’s Parente also noted CFOs should also concentrate on finding compelling use cases before deploying AI, noting that one CFO he spoke to wanted to implement a solution across the entire business, but hadn’t come up with particular applications. He contrasted this with a finance chief at an agricultural company who had a more specific plan to use the technology to determine if a cow was milked properly. 

CFOs also need to put AI products through the regular fiscally rigorous tests to determine if they’re worth their cost and fit into their budget.

“What is the true cost of this? Everyone thinks of the benefits: can I reduce my labor, improve my quality. But… what is the cost to acquire it, to build it, to maintain it? That is a financial executive’s responsibility,” Parente said. 

Controllers to review specific applications

In contrast, someone at the controller, treasurer or CAO level is more in the weeds and so their AI skill sets should revolve around the specific needs of their organization and how they could be improved, said Armanino’s Laos. 

“It becomes not so much planning but looking at practical applications,” he said for executives in those roles.

At this level, AI skills become more about specific applications, like using AI for automating compliance and reporting, error detection, and cash flow forecasting, said DeepSky’s Hsu. Looking at use cases in the financial departments for AI may require controllers and treasurers to consider how AI might disrupt segregation of duties, Laos said. 

A core principle of strong system of financial controls, segregation of duties comprises systems that are set up so that no single person is responsible for all the stages of the financial process, which can range from payment authorization and approvals to reconciliation, as defined by the University of Oxford. 

“Traditionally, roles like [the] controller and treasurer have been kept separate to prevent fraud,” Laos said. “A question worth exploring is whether AI can [serve] in dual capacities without human bias or the risk of fraud.”  

Getting skilled

Developing AI skills is a combination of learning-by-doing, consuming content on AI, and finding learning communities or events around AI, the experts said. 

Both Armanino’s Laos and KPMG’s Parente were skeptical of generic AI classes, which tend to just scratch the surface. Instead, Parente recommends role-specific training that concentrates on particular positions like the CFO, CAO or treasurer. 

Large accounting firms such as KPMG are partnering with various universities to offer such focused training. But, regardless of what one decides to do, Parente said the important thing is to get started and to not be intimidated, noting that anecdotally, some executives’ understanding of the technology remains relatively “light” as they are fearful of its complexity.

“There’s this stigma…but at some point this becomes business as usual,” Parente said. “We all learned spreadsheets, we all learned financial systems, all these things came over time. That will happen again. This is the beginning of a journey, but it’s not the first time you’ve been on it.”