Artificial intelligence (AI) is what all those 1980s killer robot movies were trying to warn us about…right?

Not exactly.


For financial institutions (FIs), AI has many beneficial aspects. With the right platform and proper optimization, AI can enhance the experience for both the institution and the customer. From credit risk monitoring to customer behavior predictions and everything in between, AI solutions can provide services that were lacking in the pre-pandemic world.

Thanks to the accelerated digitization sparked by COVID-19, those killer robots may be on our side, at least when it comes to assessing credit portfolios.


AI: Why it’s beneficial to FIs

Like a toddler after a long day at the park, COVID-19 is finally starting to wind down; just like that toddler, the pandemic left a giant mess behind. Over the past year, there have been record high levels of unemployment, and industries, particularly hospitality, have faced major challenges. Due to the unusual circumstances, there is a cloudiness around the reported numbers relating to these economic hardships.

“And that’s an area that it’s really important to consider artificial intelligence on—your numbers—because it takes the learnings that come from your portfolio and translates them into future events,” said Brian Riley, Director of Credit Advisory Service at Mercator Advisory Group. According to information published by the Federal Reserve Bank, current loss rates are 2.64%. However, in Q2 2020, credit loss rates were 125 basis points higher (3.85%), meaning that a significant number of customers had a charge off on their account.

By using AI, financial institutions can begin to address the weaker points in their clients’ portfolios. Prior to the pandemic, many of the strategies used to assess credit files were executed manually, but a lot of the pre-pandemic metrics did not compute during a time of crisis. “The payment brands are top notch and handling large volumes of data and analyzing it and doing things with it,” added Riley. “But with Brighterion, [FIs] can use that in a practical basis to look at where your portfolio is going.”

With the new strategies available through AI and machine learning (ML), a credit manager can put countermeasures in place against delinquency and other credit risk factors on a case-by-case basis. “The takeaway there is that when you look at artificial intelligence, the timing is right to deploy it in your institution,” concluded Riley.

Even if a financial institution is already using AI, it is always beneficial to test the AI strategy of that institution against others to see how it stacks up. With AI, the opportunities for growth are endless.


How AI is used for optimization

If you have a small amount of data elements and a fixed outcome for optimization, a human can probably handle that. But let’s talk about the toddler again.

Before the park, the toddler had explosive amounts of energy. He was throwing Legos, jumping on couches, and demanding the babysitter put on his favorite show or he’d scream. All that energy can be hard for one person to control, and excessive data is like a screaming toddler.

It is impossible for humans to digest large amounts of data sets, let alone process them. If FIs want to compete in the increasingly digital payments environment, they are going to need the help of AI to get that data under control, collecting and processing any and all available data.

“In many cases, you’re trying to now optimize many different outcomes,” explained Sudhir Jha, SVP and Head of Brighterion at Mastercard. “[Sometimes] you’re optimizing the default rate, sometimes you’re optimizing the profitability per customer, sometimes you’re optimizing increase[d] credit limits for certain individuals. And so there are very different sorts of things that you want to optimize.” It is imperative that the AI models are good at catering to those specific outcomes.


Considering adoption?

Let’s not think about the toddler for this one.

Each day, more and more research is conducted to create sophisticated AI algorithms. With that research, these algorithms become easier to explain. “Making sure that we can provide a very good explanation of all the things that [Brighterion’s] model is predicting is critical for adoption,” assured Jha. AI adoption, while nuanced, does not necessarily mean it’s overly complicated.

The same knowledge can be applied to the cost. Many FIs are concerned with the price tag on AI models and believe many data scientists are needed to properly implement AI solutions. While this methodology used to hold some truth, many of the platforms available today have solutions that are nearly ready to be implemented with little change. Brighterion builds custom models for their customers, using their platform, in very little time.

The other aspect of this is the technology continues to evolve at a rapid pace,” added Tim Sloane, VP of Payments Innovation at Mercator Advisory Group. “So we also see that what might be a legitimate problem today may not be a legitimate problem in two months, six months down the road.”

Now is the time for FIs to adopt the newest technologies based on the needs of their business and customer base. AI is an ever-evolving concept that shows no signs of slowing down, both in the payments world and otherwise. With the digitization and customer expectations that were both sped up and enhanced, respectively, since the beginning of the pandemic, AI is quickly becoming a necessity across all platforms.