If every CFO you knew implemented AI into their workflow, would you? It’s not just a hypothetical, like the well-worn bridge everyone’s parents used as an example. AI and finance experts think the time to educate yourself about AI is now—but not because everyone’s doing it.
Within the next few years, any finance professional who doesn’t understand the fundamentals of AI, ChatGPT, or large language models and machine learning in general, “might as well be shaking up one of those Magic 8 balls,” Glenn Hopper, CFO and director at Eventus Advisory Group and author of Deep Finance: Corporate Finance in the Information Age, told CFO Brew.“You can’t trust something you don’t understand.”
That’s why now is the time to learn. There’ll never be an easier time to learn the fundamentals of AI, he explained, because the technology will only get more complex from here. That’s not to say that you have to suddenly “dump all of [your] proprietary company data into some public large language model,’” he clarified. “But because [AI is] coming so quickly, we need to pause and understand the underlying technology and fundamentals of it.”
While Hopper’s written an entire book about the intersection of AI and finance, the core of his expertise is refreshingly simple. “I’m just a regular finance guy, but I’ve found that I can get better at my job through automation, [and] state-of-the-art automation today is artificial intelligence and machine learning,” he explained.
And that’s what he wants other CFOs to figure out how to do as well. “Finance people are—and should be—risk averse,” Hopper said. “I’m not suggesting that we should be anything other than risk averse. However, there would be a greater risk if we were to ignore [AI]. It would be like ignoring the internet in the early days.”
But before anyone jumps on board with AI, here’s a handy checklist.
Get your bearings. Step one? Understand what’s out there. Finance professionals need to comprehend the absolute basics of AI—what it is, what it means for finance, how it’s used—before they can even think of connecting the dots between the technology and their own workflows, Hopper said. Plenty of leaders have incorporated AI seemingly without basic knowledge about its functions, so, shockingly, this is actually a critical first step.
After brushing up, you need to understand “your processes as they exist today in the real world,” Hopper explained. Think of it like a quality audit of your own workflow. “You just go and look at: This is our customer onboarding, [and] go through the whole process. This is our invoicing and this is our AP, this is our AR,” Hopper said. “This is how we do all these things.”
Once you have a deep sense of the fundamentals of your actual workflow, then you can start looking for solutions. For example, maybe the way you were handling accounts payable was inefficient, so first look for fixes using your current systems. When you start running into problems without solutions, that’s when it’s time to think about new technology and automation, Hopper said.
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.
Plenty of problems could require AI for efficient fixing—but many won’t. “AI is not something you can just throw at everything and it’s a magic, curative fix,” Hopper stressed. “[Solutions] could be as simple as either changing an actual process or putting in other software for automation, but you can’t figure that out [until] you know your current landscape.”
Data literacy. Once you start to automate some processes, you’ll quickly start accumulating more data, Hopper pointed out.
“If you’re collecting customer contact information in your CRM, your ERP, [or] whatever different systems, say, ‘This system is going to be our source of truth, and we’re going to standardize all the data that we have,’” he suggested. “We build the data lexicon. We clean our data; we share the data for the company.”
That all serves a “big, important step” in any AI implementation plan, in Hopper’s telling: becoming “a data-literate company.” Eventually, there will likely be enterprise-level generative AI tools that keep data in a protected environment, Hopper said. For now, though, no one should upload any kind of remotely proprietary data on an open public site like Bard or ChatGPT.
As CFOs and other finance professionals poke around, “it’s really about using dummy data, using public filings” to experiment with the technology so that once enterprise-level tools come out, you’ll be able to use them “within a SOC 2-compliant environment,” Hopper explained.
“Once you’ve moved up on your data literacy and your data maturity, then you can start looking at AI,” Hopper said. “If you don’t have a handle on your data, you shouldn’t even be thinking about AI at that early stage.”
Baby steps. After you have a better handle on your data, then you can “start looking at incorporating some small machine learning projects” as pilot programs and test projects, Hopper maintained. From there, it becomes a matter of upskilling and reskilling your current teams, partnering with IT, and handling change management.
Perhaps most crucially, any AI implementation plan is also simply a matter of recalibrating your definition of change. While “digital transformation” has been a corporate buzzword for decades now, Hopper thinks “digital evolution” is more apt for our current era, since everyone is constantly adapting to moving targets.
“Technology is always going to be changing,” he said. “You can’t just say, ‘We did that. Now we can rest on our laurels and go on about our business.’ It’s more of an evolution, reacting to what’s out there.”