As the role of the CFO evolves, finance chiefs grapple with the prospect of incorporating AI and predictive analytics to inform decision making.
For finance leaders, however, these technologies still come with questions attached: how AI arrives at its predictions is a sticking point when it comes to generating trust in the tool, said Hari Sankar, group vice president at Oracle.
Finance professionals have always taken a “trust but verify, or maybe don’t trust and verify” mentality when it comes to their data, a mindset they have continued to apply to AI and related technologies, he said in an interview.
CFOs and members of the finance function traditionally “don’t like black box models,” Sankar said. Finance leaders also have less familiarity with the technology than those in other business areas such as marketing, who have had time to get used to AI’s quirks, he said.
“Finance, if you cannot explain why the model came up with prediction it did, you’re less likely to trust it,” he said.
Prying open that black box model is why Oracle itself is investing in technologies surrounding what the company terms “explainability,” he said. When offering predictions, for example, “we typically give you some explanation of the underlying factors and how these factors are weighted or how important these factors are in the specific prediction that we're giving you” to craft greater trust and confidence, he said.
“They may be at a different point five years from now, but at this stage of the game, without some level of transparency, finance is less likely to sort of run with the prediction that came out of the model,” he said.
Sharpening your data skills
With AI interest exploding, establishing trust in the technology is going to be critical both for CFOs who want to put it to use and to vendors of such tools that want to stay ahead in a competitive environment.
Seventy-three percent of companies reported they are prioritizing AI over their other digital investments, according to a recent survey by Accenture. Furthermore, 90% of business leaders are looking to apply the technology to “aspects of operational resilience,” the survey found, which includes data-driven capabilities, such as finance (89%) and supply chain (88%).
While experimentation — especially with generative AI tools — is growing, many finance leaders are evaluating the technology’s worth on a case-by-case basis, Sankar said.
In some areas, if machine-driven forecasts are shown to be more accurate, the firm will lean toward using the technology. In other areas, however, companies are still relying upon human forecasts if they still don’t trust that AI will be able to provide one accurately.
“But in every one of those cases, the more data you can give them, and the more input you can give them to say why the machine is likely to be better, or the machine model is likely to be well informed, I think it will get more trust,” Sankar said.
With the focus on AI intensifying, the role of data skills within the finance function is becoming more highly prized: CFOs may be more inclined to trust someone who can talk about data science or machine learning in finance-centric terms, Sankar said.
“Some of the younger hires over the last 10 years, they bring a different level of comfort with these technologies into the finance function, so they are sort of the catalysts driving change within finance,” he said.
However, “I think it's unreasonable to expect finance people to become data science experts,” he said, noting this is where vendors such as Oracle can step in.
“The level of expertise, level of comfort around these technologies is growing within finance, but I think vendors have a role to play in terms of simplifying how to use these technologies, without a lot of expertise in the technology,” he said.