As 2020 gets into full swing, artificial intelligence (AI) is in a peculiar position. It’s like that cool new kid everyone knows is coming to the party but has yet to show up.
You can bet on its arrival during the 2020s though — at least according to the message sent as part of the latest PYMNTS Masterclass with Brighterion, an AI-focused company owned by Mastercard. During the session, Karen Webster talked with Amyn Dhala, vice president, global product management, AI Express, Mastercard, about what it will take to get AI into more financial institutions (FIs) and other operations in the coming years.
PYMNTS/Brighterion research has documented the challenge ahead. Take just a few data points: 63.6 percent of FIs believe AI is an effective tool for stopping fraud before it happens, and 80 percent of AI-using fraud specialists believe the technology could reduce payments fraud. But AI is still relatively rare in the banking world, with only 5.5 percent of banks in our survey equipped with genuine AI systems. Meanwhile, the rest rely on automation and machine learning (ML) technologies.
In other words, FIs that use AI are few and far between, but that’s not to say FIs aren’t investing in it — or rather, in what they think it is.
In fact, the question of where to begin with AI is a big challenge going into 2020, Dhala told Webster.
“They don't know exactly where they should be starting from,” he said. “As you move along, there are other customers who want to actually develop a business case for leveraging AI. ... Then on the other end of the spectrum, you have customers who are already using AI and are looking to leverage and add the state-of-the-art AI technology to make continuous improvements.”
Value Of Speed
One key to getting over the starting line for AI? Speed.
Earlier this year, for example, Brighterion launched a program called AI Express. Its goal: to build “a fully functional [AI] model, customized to your unique business situation,” within five to eight weeks.
“In terms of developing an ROI, that's what AI Express is all about,” Dhala told Webster during the PYMNTS Masterclass.
Speed can also offer opportunities for experimentation when it comes to AI and the use cases tied to the emerging technology.
“AI Express is an excellent way for you actually to check out new product propositions,” Dhala said. “We do see customers trying to see in new areas, and unique areas, how they can actually leverage AI. In fact, we make use of such similar approaches, with Mastercard as well, wherein we look at doing AI-based initiatives in our quick runs, to see how we can … potentially deploy those products for our customers.”
The general idea is to identify and create AI-enabled use cases, which in turn — in a kind of virtuous cycle — could further spark AI innovation and more deployments. According to Dhala, those early use cases for the payments and financial services realm are becoming clear — echoing the data from the PYMNTS research report and reflecting larger economic trends, such as rising loan delinquencies in some areas.
Core use cases that are getting a lot of traction, Dhala said, involve credit risk.
“It’s something that is gaining prominence across the globe, given the current economic climate,” he noted. “Any marginal improvement in terms of modeling or accuracy can result in significant gains because there’s a reduction in credit losses. There are not too many other areas that deliver those kinds of gains, which have an impact in terms of profitability. There’s that particular competence. Organizations would want to leverage AI and see what benefits they could get from that.”
Dhala dug deeper into that point.
“In terms of credit risk, we’ve seen benefits wherein we are able to identify incremental delinquencies of up to 20 percent, based on portfolios, without impacting the customer experience,” he said. “When you start looking into that from a broader portfolio perspective, we are looking at a significant impact in terms of customer profitability and a better customer experience.”
AI can also help to spot credit risk.
“When we work with our customers in terms of identifying what data sources we could use, we look at sources such as transaction data,” Dhala noted. “We look beyond the historical data sources that are currently being used — obviously bureau scores, which may be utilized. But we're also looking at other aspects, such as transaction data and bill payment history, and other aspects such as log inactivity.”
That’s not all.
“Apart from that, from a customer management perspective, customer attrition is another use case that is gaining prominence,” said Dhala.
The rising importance of personalized customer experiences in commerce is also sure to spark AI innovation and deployment.
“We have product recommendations or next-best offers,” Dhala noted. “The intent is to provide a hyper-personalized experience to the customer, with what is right for the customer at that point in time.”
Another likely early focus of AI is use cases that protect customers from rejection due to being wrongfully perceived as a fraud risk.
“Through AI, you can actually reduce false positives, which in turn helps reduce customer attrition,” Dhala told Webster.
The end of the year or the beginning of a new one is always a good time to take stock and make an action plan. That certainly holds true when it comes to AI, according to this latest PYMNTS Masterclass. The time is now for AI to be baked into business plans, which was one of the main messages to come out of the session with Dhala and Webster.
“Stop putting it into your roadmap for 2021,” Dhala said. “It should be on your roadmap for 2020. Get started, get the roadmap — because if you don’t, you’re going to get left behind.”