It’s always been prudent for finance and IT to work together. But now that AI and especially generative AI are here, the two functions are becoming intertwined like never before.
Some CFOs still would prefer to stay in their silos, Russ Blattner, cofounder and CEO of AI governance platform Superwise, told CFO Brew, and only deal with the budget aspects of technology. That’s a mistake, he believes.
In the AI era, “the CFO is going to have to become more of a pitcher than a catcher,” he said. They’ll need to actively work with CIOs rather than “just sitting back and receiving proposals for spends that include AI.”
With ubiquitous AI comes ubiquitous risks: That’s because AI is (1) everywhere, and (2) really, really risky. It’s becoming “integrated into everything,” Blattner pointed out. An enterprise organization can use “up to a thousand” different types of software, he said. If only half of those add AI, that’s still 500 types of AI the company will need to contend with. If your video conferencing platform starts using AI to transcribe and summarize recordings, for instance, you’ll want to know what the platform’s doing with that data, where it’s being stored, and whether it introduces new security risks.
AI is not only amplifying existing risks but also introducing new ones. Cybercriminals can “inject” data into AIs or other systems to influence their behavior, Blattner said. For instance, fraudsters could make it more likely that a bank’s AI would approve them for a loan they plan to default on.
Then there’s drift, or the tendency of algorithms to degrade or veer off course over time. One of Blattner’s client’s rewards programs, for instance, drifted and started “giving out massive rewards.” His team detected and reported the problem and the client shut it down within a day.
News built for finance pros
CFO Brew helps finance pros navigate their roles with insights into risk management, compliance, and strategy through our newsletter, virtual events, and digital guides.
But undetected drift can have serious financial consequences. In 2021, real estate platform Zillow lost 45% of its market cap and had to lay off 25% of its workforce after it bought too many houses that its Zillow Offers algorithm had overvalued. The algorithm had likely failed to adapt to changing market conditions.
And that’s not even getting into runaway agents, or the risks that can be introduced when employees use AI to create tools on their own.
CFOs need to be more proactive about AI safety, Blatter believes. They can’t just take a “let me know when something happened” approach, he said.
From R&D to ROI: But though AI certainly comes with its share of terrifying possibilities, it also brings opportunities. CFOs should know that AI is maturing from its R&D era into its ROI era, Blattner said. “For years, we were telling [companies] ‘just trust, you’ve just got to invest, in a couple of years you’ll start to see the results,’” he said. Now, he said, those results are visible. “You should be able to see ROI [from AI] within three to six months, easily within a fiscal year,” he said.
CFOs need to be tracking this ROI “in lockstep” with their CIOs, Blattner said. The CIO can put systems in place to track AI-related metrics, he said. but will need the CFO to interpret “the impact on the business and potentially the revenue side of operational efficiencies” that are created.