Why banks are using AI for risk management

With its data-heavy and often time-intensive processes, risk management is one area where community banks may be eager to put artificial intelligence to good use. Community bankers and AI experts explain how they’re using AI-driven regtech solutions.

For many community banks, it’s still early days for using artificial intelligence (AI) to understand data and improve risk management and regulatory compliance processes. But that doesn’t mean they’re not intrigued.

Mickey Marshall, director of regulatory and legal affairs at ICBA, says many community bankers are becoming interested in AI because it may help them solve a few nagging compliance problems or make data-intensive processes more efficient. He notes that until quite recently, when community bankers heard about AI, some immediately thought of HAL 9000 from 2001: A Space Odyssey.

That’s changing. Marshall is convinced that community bankers are beginning to understand how the latest AI technologies can aid in rote processes, allowing them to devote more time and energy to maintaining relationships in highly personal ways.

Brice Luetkemeyer, president and CEO of $185 million-asset Bank of St. Elizabeth in St. Elizabeth, Mo., has begun exploring how AI might improve the community bank’s risk management processes and marketing to customers.

The community bank is mulling how to use this technology for Bank Secrecy Act (BSA) compliance and other risk management processes. Luetkemeyer says community banks should work with regtech vendors as an educational exercise to make sure they’re evolving to meet the demands of a diverse clientele. For example, Bank of St. Elizabeth works with Neocova, a St. Louis-based data platform and solutions provider with a specialty in AI technology.

“Neocova and other regtech companies can help give you a broader grasp of what AI really means,” Luetkemeyer says. “The more you understand AI, the further you’ll go with it.”

One problem in understanding AI is that the term itself can be something of a catch-all. “AI is math and statistics,” says Matt Beecher, CEO of Neocova. “It’s RPA [robotic processing automation]. It’s predictive analytics. It’s machine learning and deep learning. And it’s all of these things rolled into one.”

Beecher believes that a good way to understand the AI’s effectiveness toward increasing customer engagement is to look at a core function like BSA and anti-money laundering (AML) compliance. He notes that complying with BSA/AML requirements generally means dealing with high rates of false positives, possibly 90% or more. By dramatically reducing these false positive results, bank employees can focus on and manually review only the most likely infractions.

Seizing AI opportunities

In what turned out to be perfect timing, Cross River Bank began investing more in AI and machine-learning technology in late 2019, “just in time to deploy these techniques across PPP loans originations,” says Jesse Honigberg, technology chief of staff at the $13.5 billion-asset community bank in Fort Lee, N.J.

Cross River Bank’s $13 billion in PPP loans was possible, in part, because of AI technology that allowed the bank to better understand borrower documents and identify risk. Honigberg says by having machines flag potential problems, the bank knew when it needed to do a further manual review. “If A plus B didn’t equal C, then we knew it had to be looked at again,” he adds.

Working this way led to greater efficiencies. “Often, one person could do the work that four or five would have done in the past,” Honigberg adds. “Our employees can be much more directed, as opposed to trying to find a needle in a haystack.”

Cross River Bank teamed up with Ocrolus and a number of other vendors specializing in AI and machine learning to begin gathering data and flagging documents for possible problems. While fintech and regtech companies use AI to help their community bank customers automate compliance and mitigate risk, many community bankers are adding their own twist to the solutions that vendors supply. Honigberg says, for instance, that his bank applied its own algorithms to those of its vendors.

Sci-fi no more

While AI has clear implications for rote processes, it can help with more nuanced issues, too.

For example, Yogesh Pandit, CEO and founder of New York City-based Hexanika, says AI can make regulatory compliance less reactive and more proactive.

Too often, Pandit says, community banks submit reports to the CFPB at the end of the year without knowing whether the loans they’ve made are consistent with prevailing fair lending practices. Hexanika helps banks analyze data to know what a pattern of fair lending in their community would look like, so they can proactively tailor their lending practices and make loans accordingly.

The days of associating AI with science fiction are numbered. Many community bankers are coming to realize that as compliance burdens rise, understanding data and using technology to help make regulatory compliance more efficient can be a game-changer.

Luetkemeyer, for instance, is convinced that the “old, static way of thinking, where banks produce simple reports off their cores,” has got to change now that AI has opened up a wealth of new possibilities. “If banks don’t move in this new direction,” he adds, “they will be missing out.”

Originally Posted to: