NEW YORK CITY — The use of artificial intelligence-based risk modeling is far from foolproof in a recessionary environment.

While advancements in machine learning (ML) capabilities and data integrations have dramatically increased the scale and complexity of AI-based risk modeling, the models are only as good as the ways in which they are employed, Pankaj Kulshreshtha, founder and chief executive at lending automation firm Scienaptic AI, said Wednesday during a panel discussion at Fintech Nexus 2022 in New York City.

“How you handle a different macroeconomic environment is a totally different matter than whether you’re using more advanced machine learning models, or using historical logistic regression types of models,” Kulshreshtha said.

AI-based models are “fine” for recessionary environments, if cut-offs and metrics are properly adjusted, Kulshreshtha added.

The $11.4 billion Bethpage Federal Credit Union, a Scienaptic AI client, was “comforted” to see Scienaptic AI’s modeling validated during the height of the COVID-19 pandemic, John Witterschein, vice president and head of consumer credit at the credit union, said during the panel.

The credit union did not include 2020 performance in its validation of Scienaptic’s model, due to delayed loan repayments and an influx of cash.

Paul Quintero, chief executive at SMB lender Ascendus, echoed Kulshreshtha’s thoughts on modeling. “I don’t believe that any model will shield you from a recessionary risk, because we’re all in that same water, right? Boats are going to rise and fall,” Quintero said.

“I think we have to recognize that there are other ways to mitigate those kinds of risks. I’m not sure a model will help you avoid a recession.”

Bank Automation Summit Fall 2022, taking place Sept. 19-20 in Seattle, is a crucial event on automation and automation technology in banking. Learn more and register for Bank Automation Summit Fall 2022.