AI in financial services has moved well beyond its early applications in fraud detection and chatbots. Financial institutions are increasingly using machine learning to analyze consumer behavior, personalize product recommendations, improve operational efficiency, and identify opportunities to engage customers more effectively. However, the success of these systems depends heavily on the quality and accessibility of the underlying data.
Collecting, organizing, and cleansing enormous datasets can be one of the most challenging aspects of implementing AI. As machine learning capabilities advance, financial institutions are also finding new ways to automate these data-management processes, potentially expanding the number of practical AI applications throughout the organization.
Despite articles to the contrary Artificial Intelligence can’t predict everything. This article steps through the many places AI is being used in financial services to predict consumer behavior. What is doesn’t discuss is how difficult it is to collect and clean the data required to feed the machine learning tools or how new machine learning tools are now being used to assist in accessing that data and cleansing it. :
“That can include suggestions for payment cards, marketing and other emerging payment and financial product verticals that can be delivered in real-time. The recommendations aren’t direct loyalty programs per se, but can be part of a recommendation that includes a number of factors that accompany a payment program, card or financial service.
‘The use of data has always been important as a way to manage programs,’ Murray said. ‘It’s the magnitude of data that’s available that’s enabling the learning.’
Artificial intelligence has become increasingly popular among financial institutions and payment companies over the past two years, given its ability to manage and analyze large amounts of data, a practice that improves over time as more data accumulates about specific uses, accounts and relationships with different companies.
AI has found a welcome use case in managing security risk. Citigroup, for example, has partnered with Feedzai to match new payments with past transaction records to quickly spot possible errors or fraud. AI is also being used to thwart digital attacks on payment systems.
There is a short diversion discussing bots and then back to new use cases with a statement from an analyst suggesting that the two primary use cases for AI is fraud and chat bots.
In 2017 I authored the report “Now Is the Time to Develop an AI Business Plan Beyond Fraud” and my new report “70+ Processes Banks Have Already Improved Using AI” will be published this month which dispels the myth that AI is contained to only a few areas within FIs. In fact, if there are financial institutions that have not yet assigned a team to analyze where AI should be used beyond fraud detection to improve operational efficiency and better engage customers it will almost certainly need to scramble to catch up:
“Additionally, more creative uses for AI are emerging, such as voice controlled payments and in-car shopping based on a rider’s travel route.
‘If you think about natural language processing and connections with consumers in real-time and couple that with AI techniques such as machine learning, you can get to a point where you can take recommendations to consumers to another level,’ said Tiffani Montez, a senior analyst at Aite Group.
As consumers answer questions about rewards or card preferences, an AI-driven analysis can take place in the background on how that consumer spends money to determine whether a miles-based perk, cash-back or other travel rewards would best fit that particular consumer, Montez said, adding this can make preapprovals potentially more effective because there is more analysis of backward looking payments along with current data.
“It’s early yet for that kind of use because most of the adoption of AI has been in two directions. Lots of people are using it to power chat bots or interactive assistants, while others are using it for AML and fraud,” Montez said.”
AI in financial services should not be viewed as a technology limited to fraud prevention, AML, or customer-service bots. Applications ranging from personalized card recommendations and real-time marketing to natural language processing and operational automation demonstrate how broadly machine learning can be deployed across financial institutions.
The competitive advantage may increasingly belong to institutions that identify where AI can improve processes and customer interactions throughout the enterprise. Financial institutions that develop a comprehensive AI strategy—and address the underlying challenges of accessing and preparing their data—will be better positioned to take advantage of new applications as the technology evolves.
Overview by Tim Sloane, VP, Payments Innovation at Mercator Advisory Group
