Trustworthy AI helps Regions Bank better serve customers

Financial institutions worldwide are feeling the scrutiny from both customers and regulators alike. Perceptions of an institution’s governance practices, including its commitment to ethics, fairness, explainability and transparency of decisions, are critical to its standing. No wonder those poised to gain a competitive advantage today want to ensure their AI is fair, trustworthy, and explainable.

A member of the S&P 500 Index, Regions Financial Corporation is one of the United States’ largest full-service providers of consumer and commercial banking, wealth management and mortgage products and services. This Birmingham, Alabama-based organization has extended its culture of doing the right thing to both its customer relationships and its approach to AI.

“Trustworthy, transparent models are critical to our success and really go back to our culture and key tenets — “to serve our customers,”

As banks, insurance companies and other financial institutions look to innovate with AI, the new currency is trust. Although the use of artificial intelligence continues to grow across industries including financial services, trust is at a premium and that’s bringing greater scrutiny on AI deployments, according to IBM and Morning Consult’s Global AI Adoption Index 2021. More importantly, the index reveals that 91 percent of businesses using AI say their ability to explain how it arrived at a decision is critical.

Trustworthy AI requires data completeness, accuracy and quality, and the data underlying the models must be representative of the data used to make the decisions. Plus, the models must be “explainable,” meaning their decision-making processes are easily understood. This is especially critical in the highly regulated world of financial services.

Regions wanted to create a trustworthy framework for AI that included ModelOps capabilities and the ability to identify data and model drift. That meant that it needed tools and processes to monitor data drift and ways to ensure models could be adapted if the data started to change. Misra and his team worked with IBM Data and AI Expert Labs and the IBM Data Science and AI Elite team to align with data tools, methodology and personnel. Part of this effort involved understanding how IBM Cloud Pak® for Data could help them assess data drift, measure model performance, and keep their personnel informed.

Read here about the methodology they used to develop high quality and trusted AI.

When Misra joined Regions, it was critical to demonstrate the value that data and AI could bring to the business. Rather than starting with a small project, he looked to make the biggest impact quickly.

“I had to make sure that we could show that we could move the needle and deliver large amounts of value to the business,” he said.The first data project Regions built delivered tens of millions of dollars to the business in additional revenue while saving losses. “I used that as a way to demonstrate to other parts of the business: ‘look, we’ve done this, we can do this for you as well.’”

Soon, there was more demand than Misra’s team could meet. “It was something they signed on to and became big proponents of, so much so, that innovating with digital and data is one of three strategic initiatives for the bank right now.”

Misra explained that to create trust in business decisions driven by artificial intelligence, a variety of stakeholders in the second and third line of defense provide oversight into the quality of the company’s models. The result has been trusted data products (including those that help reduce fraud for the bank, assist commercial banker and wealth advisors, and provide insights into consumers) so Regions can better serve customers.