Banks must focus on the human equation in AI, machine learning
Banks are making big investments in artificial intelligence (AI) and machine-learning software to improve the customer experience and detect fraud. AI-enabled chatbots and voice assistants are already the norm at major banks.
Financial institutions are also employing AI-based software to make lending decisions. Recently, Visa unveiled an advanced AI-based transaction-processing system to approve and decline credit and debit transactions on behalf of banks in the event of a network outage. Banks are expected to spend more than $7 billion on AI this year and as much as $14.5 billion by 2024, according to a recent article in The Wall Street Journal.
Despite the enormous benefits AI can deliver, financial services companies should take note of its drawbacks. Several widely reported blunders committed by well-known companies serve as prime examples.
One was when a Minneapolis man discovered his teenage daughter was pregnant after Target’s predictive algorithms started sending coupons addressed to her for baby items. Equally memorable was Uber’s surge pricing in the early hours of New Year’s Day when revelers were charged as much as $1,113 for a 21 minute ride. More recently, regulators are investigating Apple Card for gender discrimination after customers accused the credit card’s algorithms for extending higher credit limits to men.
These examples underscore AI’s failure to capture human judgment. Uber’s surge pricing makes perfect sense to the laws of supply and demand. However, the ride-sharing firm neglected to consider how humans perceive price fairness. While Target’s algorithms ensured the retailer could deliver relevant offers to customers, they did not consider the emotional dimension.
Banks have to become more emotionally intelligent and recognize their customer’s cognitive, cultural, emotional, psychological and social dimensions. They have to be more attuned to customer pain points and aware of junctures in the customer journey that require empathy, human judgment and clear communication. As AI use rises, the human side of the equation becomes more critical and more non-negotiable.
Address customer pain points to build relationships
Instead of deploying to AI to invent the next “big thing,” banks should use it to alleviate customers’ existing pain points, such as unusual transactions or overdraft fees, and thereby increase customer loyalty.
Recently, a senior advisor at Simon-Kucher was pleasantly surprised when her credit card issuer flagged one of her charges. She had ordered takeout from her favorite restaurant, where a $10 tip was added onto a $30 bill, instead of her usual $6 tip. The bank’s algorithms detected the unusually high tip and alerted her accordingly. It was a small amount, but went a long way to earn her loyalty for the bank.
Using an AI chatbot to anticipate a customer’s reason for placing a customer call can also improve loyalty. One Citibank customer reports being properly impressed when the bank’s AI chatbot was able to quickly narrow down the reason for his call and automatically waived an overdraft fee. The customer must be at the center of the equation.
Another appropriate application of AI and machine learning can be to improve self-service channels and make it easier for customers to perform basic online banking transactions, like making payments, managing finances or opening an account. Finance and bank processes can be complex and confusing. Forms can be tedious to fill out. Payments can be difficult to set up. AI can be programmed to observe online behavior to detect when a customer is having trouble navigating the website or filling out a form. Providing a chat box with a live assistant helps the customer navigate these tasks more efficiently.
AI and machine learning also have extensive uses in marketing to improve targeting models, personalize messaging, manage real-time programmatic advertising and so on.
However, during these turbulent times, past performance is less of a predictor of future success. Most targeting models, product mix, offers and entire marketing plans are based on what has worked in the past. How do we use past data to make prudent market decisions in the midst of uncertainty? How do we evaluate multiple scenarios? The right answer is not to throw away the data and resort to human intuition alone, but instead, use augmented intelligence.
Augmented intelligence is a more robust approach that incorporates the domain of human knowledge with the best of AI. A host of startups are finding ways to incorporate human intuition and judgment in AI to automate marketing and sales. Here we are looking at a human-centered approach where AI plays an assistive role to enhance, rather than replace human intelligence.