BACKGROUND IMAGE: Ayvengo/stock.adobe.com
It has long been a dream of many CFOs to have within arm's reach a crack team of financial analysts who can provide rich insights on trends and possibilities hiding deep in data. Well, the unimaginable might finally be imaginable.
Artificial intelligence -- or, at this moment, the foundational layers of AI -- holds the promise of extracting the kind of valuable financial information that will set companies free to confidently take chances and explore new opportunities. That promise has been foreseen by tech industry analysts and executives who believe that, while the first chapters of AI for finance are just now being written, it's nonetheless the right time to start looking at a technology that could revolutionize finance.
"The ability of machines to crunch numbers is phenomenal," said Leon Katsnelson, director and CTO of analytic platform emerging technologies at IBM. "But they still try to emulate a gut feel, and it's not really a gut feel. In a way, cognitive learning is a way to build up gut feel. And you don't have to be a math scientist wearing a white lab coat to use this technology."
Advocates of cognitive learning technology ascribe that computers think at a level close to human cognition, as IBM advertises with its Watson technology. But some experts claim that cognitive learning using machines is a myth of marketing, and that it's not possible -- at least, not yet.
AI works in other ways, however. It also includes machine learning technologies that use algorithms to improve computational capabilities as more data is obtained. It's an area of artificial intelligence that few deny is popping up in mainstream business and with financial applications.
"The availability of data and information inside companies is breaking silos and transforming the world of finance," said Frederic Laluyaux, CEO of Aera Technology, a business intelligence software vendor.
Prior to AI, companies could crunch numbers and see a version of truth, but would ultimately need a financial analyst to maximize the information, Laluyaux said. But with data becoming cleaner and more reliable, algorithms are now making smart predictions that businesses can use with confidence.
Early examples of using AI for finance
While AI has been around for years, a confluence of elements is making it possible for industries such as banking, insurance and wealth management to use it in a variety of ways to supplement or replace human labor, including reviewing transactional patterns for fraud or powering chatbots to communicate with customers. Cloud computing, lightning-fast processing speeds and big data have enabled enterprises and even small businesses to tap into these and other AI-type technologies.
Diego Lo Giudice, a Forrester analyst, cautions that the lines of AI aren't easy to draw. He and others believe basic, traditional machine learning that uses algorithms on structured data is not AI, whereas the use of that technology to make sense of unstructured data can be considered a step toward AI.
"There are tactical and transformative parts to AI, and now would be a good time for enterprises to get into it," Lo Giudice said. "It can crack some hard problems. It requires skills and some learning, but if you start experimenting with this technology -- with data scientists or AI technologists that can be hired -- and put it on the side of business with the thinkers, it can do new things."
CFOs considering AI for finance applications can place most of these applications on top of their platforms, as long as they don't set expectations too high, Lo Giudice said.
"It's useful if you scale down expectations and start small, applying it initially to small problems," he said. "The technology is maturing every day. The research is improving and the algorithms are improving."
Laluyaux said when Aera brings new customers on board, they keep their technologies in place, while Aera "puts another skin on top of the original landscape." Aera runs customers' replicated data through its intelligence engines and returns it to its transactional layers.
"The data needs to be available, normalized, in real time, and not stuck in systems for computers to understand it," he said.
Added Laluyaux: "Algorithms and machine learning allow you to do things you couldn't do before."
Laluyaux considers this to be AI, although he acknowledges the analytical conclusions reached by the technology aren't autonomous, which is the future of AI envisioned by many.
Patience needed in the early innings
Indeed, if the development of AI technologies is analogous to a baseball game, "We're in the second inning of looking at this stuff," said R "Ray" Wang, principal analyst, founder and chairman of Constellation Research. "We've got a while to go, and there is a lot of hype out there. There's a lot of promise with the technology in predictive analytics that's now called AI. But to make AI work, a lot of stuff has to happen."
For Wang, that stuff includes the ongoing challenge for companies to coalesce all the data they collect from many sources. To even consider AI, they first need to establish clear data structures and eliminate false positives in data, he said.
But if a financial department is nonetheless considering investing in these early stages of AI for finance, Wang said IBM's Watson has "a lot of good tools," although it requires time to get it where you want it.
Offerings from Wipro Holmes, Oracle and Workday also have their advantages and disadvantages, a reminder that the technology is not yet aligned with its promise.
"[These are] the building blocks for AI. That's what people have to realize," Wang said.
Before investing in the technology, CFOs should ask what the top business outcomes they'd like to realize from using AI for finance are, Wang said.
"What are you trying to do? Start with the important business questions and see how the technology can handle those assignments."
At the least, AI has already started shifting the role of CFOs, said IBM's Katsnelson. Previously, he said, the expectation of a CFO was to keep accounts straight and be aware of potential regulatory and compliance issues.
"They focused on product rather than outcome," he said. With AI and other technologies, "there's an expectation to also be the driver of business strategy. You can't survive by being efficient only."
Get a glossary of AI terms
Learn about applications of AI in HR
Read how AI might work with ERP