Artificial intelligence is undoubtedly one of the biggest — if not the biggest — buzzwords in technology today.
From chatbots and virtual assistants to autonomous vehicles and facial recognition, the broad application of such tech is having a profound, effective change around the world in myriad fields.
And it’s effectiveness that’s making it the key to fighting fraud for financial institutions and the fintechs that help them do that, says Brendan Deakin, general manager of the U.S. market for Provenir, a software firm that offers data- and AI-powered risk decisioning solutions.
In Deakin’s and Provenir’s parlance, AI is “fit for the fraud fight.”
“Within financial services specifically, there’s just so much data that can be leveraged,” Deakin said. “It’s hard to figure out what’s most predictive and what can detect the most fraud activity over other data sources. AI really helps lenders sift through that and can quickly gain insights into which data sources are best at detecting specific suspicious activity over others.”
While some data sources may be used purposely for a particular type of lending product or a typical channel, Deakin said AI can help lenders best assess what data is most effective in assessing fraud.
“I think the real power is that AI can actually predict or detect patterns of behavior and suspicious activity that would take a lot longer for traditional methods to uncover and it can react to it very quickly,” he said.
“It can self-adapt. It learns from every transaction that flows through the model. It’s constantly getting better based on what it learns as it makes one fraud assessment after another.”
Seeking best-possible outcome
Provenir, which is headquartered 25 miles west of New York City in Parsippany, N.J., was founded in 2004 and built its business with a traditional licensed software product, moving its product resources into cloud computing in the mid-2010s.
It was around that time when Deakin and his colleagues started hearing more and more about machine learning and AI capabilities. Work to incorporate AI into its product started three-and-a-half years ago, with Provenir adding AI model support to its decisioning platform in 2021. Since then, Deakin said, Provenir has evolved to the point where they’re not just supporting their own clients, they’re in the model generation and deployment business as well.
“We’ve generated a low-to-no-code user interface that gives our clients the ability to actually build and generate these models, and then quickly get them deployed into the transaction so that they can reap the benefits for it,” he said.
“We give monitoring capabilities through charts. Our clients can see the performance of this model in real-time, so it went from us dipping our toe in AI by just supporting the capability natively on our platform to now becoming much more of a one-stop shop as it relates to folks who want to start and build and deploy these models on their own using Provenir.”
The beauty of what AI can do, according to Deakin, is in digesting a great amount of information when used in fraud detection to ensure it determines the best possible outcome from the details it’s given.
“We don’t see a lot of false positives or errors within the fraud space,” he said. “When you get into credit, explainability is important, especially if an AI-powered credit model declines a consumer. There must be reason codes for why the model decided to decline a consumer. That’s where we haven’t seen great adoption yet of AI-powered credit-risk assessment models, but we’re certainly seeing it in fraud.
“This is simply a measure of whether this transaction, this application, whatever it might be, can move forward because we are confident that it is who they say they are … That’s really where we see great promise and great results in our clients using AI for fraud.”
Read more on AI and fraud detection
- FIs Face Barriers to Fraud Fighting AI Adoption
- How new technology is changing the face of fraud detection in neobanks
- Provenir announces strong growth in 2022
