- Key Insight: Community banking experts say AI can only be useful if it's equipped with "good, clean" data.
- What's at Stake: Lenders of all sizes are using AI to improve efficiency, but big banks have far more data at their disposal than community banks.
- Expert Quote: If a bank uses low-quality data to inform its AI tools, the result is "garbage in, garbage out," said Shawn Main of Vantage Bank.
Artificial intelligence, banks are learning, is only as good as the data you feed it — and community banks know that as well as anyone.
At a webinar hosted by American Banker last week, three community bank executives discussed how their firms are incorporating AI into their operations. One point they agreed on is that plentiful, accurate data is crucial for the technology to be useful. Otherwise, as one panelist put it, what you get is "garbage in, garbage out."
"If you want AI to work at all, it's going to rely on good, clean, solid data," Shawn Main, chief business architect at Vantage Bank in San Antonio, Texas, said during the April 21
Today, banks of all sizes use generative AI for a number of functions, including fraud detection, customer service and "back-office" tasks involving large amounts of paperwork. But in order for the technology to do these things well, it must draw from an accurate pool of data — or "data lake," as Main called it — about the bank's customers, transactions, policies and many other subjects.
In this area, big banks have a built-in advantage: With their vast customer bases, they have access to far more data than smaller lenders. To stay competitive, Main said, community banks need to start building good data-collection skills now.
"That's a skill set that I think most community banks have not developed, and they need to start developing that very rapidly," Main said. "You don't want to wait 'til you're too big, until you have too much data, to try to wrangle that back in later."
Many community banks see AI's potential. In 2025, 52% of community bankers expected the technology to make a big impact in their fraud detection, according to a
Of course, big banks have also put the technology to work. At its
But before a large language model can do all that, it needs good data. Providing it is a skill that employees in every part of a bank should learn, said Aleda De Maria, chief operating officer at PeoplesBank in Holyoke, Massachusetts.
"Everybody thinks it's a behind-the-scenes-type thing," De Maria said. "But really, for the frontline teams too, there should be an interest in data, data oversight and data strategy."
Jim Kisch, CEO of Passumpsic Bank in Saint Johnsbury, Vermont, calls this process "data cleansing." Once a bank has mastered it, he said, the many benefits of a well-informed AI bot can be unlocked — including freeing up employees' time for more interesting tasks.
"We want to use AI to take more [of the] operational backroom offices and create more specialized roles, even creating new roles" for human employees, Kisch said.
At Passumpsic, those new roles may soon include a "fraud grief counselor" to work with customers who fall victim to scams, Kisch said, as well as a "new entrepreneurial role in advising new businesses."
"All those are made possible because we're shifting from AI operational roles to more direct contact in the community," Kisch said.
Main echoed that point.
"This is all about driving more efficiency in your back office so they have more time to spend with your customers," he said.
At PeoplesBank, De Maria said, AI is viewed as just one of several technological solutions. And while it works very well for some problems, it's not right for every single one — especially if the data is lacking.
"That's been a focus for us for this year, is making sure we have the data ready for us to deploy AI," De Maria said. "It doesn't make sense to deploy it if you have garbage data."
