How Financial Institutions Can Use Artificial Intelligence
From fraud detection to compliance, analytics powered by AI can make a big difference.
- by Mike Chapple
Financial institutions use artificial intelligence in a variety of ways, ranging from traditional analytics targets to innovative applications of AI technology. Those seeking to improve their analytics capability often begin by conducting a data inventory and then aligning that inventory with business requirements to identify opportunities where analytics can make a powerful difference for the business. Common use cases for AI in the financial sector include fraud detection, portfolio management, sentiment analysis, product recommendations, customer service and compliance.
The real-time analytics capabilities of artificial intelligence help financial firms improve fraud detection technology by increasing its predictive accuracy and, therefore, reducing the number of false-positive alerts generated by fraud detection systems. These false positives annoy customers, who may suddenly find their credit card cut off, and consume costly time from financial institution staff who investigate alerts and handle contact with irritated customers.
AI improves fraud detection by allowing institutions to bring massive quantities of data to bear on the problem and mining that data for the nuggets of knowledge that can confirm or refute suspicions of fraud. For example, a bank might observe that a customer is suddenly making purchases in Thailand, having never before left the United States. Traditional fraud detection techniques might immediately trigger a location-based fraud alert that would cut off the customer’s card during a critical period of international travel. AI-based fraud detection, on the other hand, might notice several events that are related, but separated in time:
- Four months ago, the customer purchased a plane ticket to Thailand.
- She checked into a Bangkok hotel yesterday.
- She accessed the bank’s smartphone app using two-factor authentication from Thailand this afternoon.
Using this information, the fraud detection algorithm may note the unusual activity, but automatically evaluate it as not likely to be fraudulent because the big picture of that customer’s activity indicates that she is traveling in Thailand.
This same technology can also be used to detect fraudulent activity that might otherwise go unnoticed. For example, if that same customer’s credit card is used at a restaurant in Virginia while she is staying in a Bangkok hotel, that may trigger a fraud alert, even if the Virginia restaurant activity would not be unusual if the customer were at home.
Artificial intelligence is also making inroads in the field of portfolio management. The past three years have seen the growth of AI-powered “robo-advisers” that move customer funds in and out of index funds and other investments based on the customer’s investment objectives, risk tolerance and market performance. The services provided by these automated investment advisers recently expanded to include dividend reinvestment, portfolio rebalancing and tax-loss harvesting capabilities. Expect to see further growth in this area as institutions continue to invest in AI and introduce automated advising services capable of making more sophisticated investment choices.
Financial institutions, like any other organization, want the ability to monitor consumer sentiment and gauge how customers react to news reports, social media coverage and other trends. Sentiment analysis techniques allow firms to monitor traditional media, social media, product reviews and other online sources, and analyze whether coverage is favorable or unfavorable. Real-time updates allow the institution’s social media team to rapidly respond to online activity, and, when appropriate, intervene before a small issue grows into a crisis.
AI solutions power product recommendations across a wide variety of industries, ranging from the “You May Also Be Interested In …” section on Amazon to the “People Like You Watched …” recommendations on Netflix. The technology powering these recommendations can also be brought to bear on financial products, by recommending credit cards, investment opportunities, insurance plans and other products to the consumers most likely to purchase them.
AI is also finding applications in the customer service space. AI-powered solutions often answer the telephone in many call centers, handling customer interactions without requiring the intervention of a human specialist. Voice recognition and sentiment detection technologies leverage AI to evaluate the performance of human representatives when they do get on the phone. These AI solutions can detect frustration, anger and other emotions in a customer’s voice and recommend appropriate interventions.
Financial institutions face a wide range of compliance obligations that limit the commitments, assertions and information that representatives may offer. AI-powered solutions can analyze the content of outbound email messages, interpret voice conversations, read text messages and evaluate other communications to identify potential compliance issues before they reach customers.
To learn more about AI solutions, download the white paper "Smart Data for Finance."