Artificial Intelligence May Actually Help Humanize Financial Services

The financial services sector has long been criticized as being insulated, elitist, and discriminatory. Will artificial intelligence finally open up and “democratize” this industry? It’s likely, and will happen a number of ways – by empowering customers, by opening up services to underserved communities, and by increasing the breadth of capabilities companies can offer.

While still in the minority, a growing number of financial services executives are bringing in AI as a part of their customer experiences and operations. About half of 500 executives (48%) responding to a survey conducted by Economist Impact and SAS in March 2022 identified advanced data analytics as among the most important technologies to harness, and 34% specifically cited AI and machine learning as their paths to the future.

Similarly, a study from Deloitte AI Institute confirms that 32% of financial services executives indicate their organizations use AI. “It’s undeniable that AI is the future of financial services,” the study’s authors state, adding that while “many FinTechs have embraced AI, the financial services industry is largely in the early stages of AI adoption.”

AI and machine learning introduce great complexities, and many financial services companies are still assessing where and how to invest in these approaches. “There are a lot of moving parts with AI and machine learning,” says Michael Upton, chief digital officer at First Tech Federal Credit Union, which exclusively services Microsoft, Amazon, Intel, Hewlett-Packard, and employees of other technology firms. Once put in place, however, these technologies have a critical role to play in the emerging digital enterprise. “Covid really accelerated digital, and the industry did a good job in customers’ needs from a tactical and transactional perspective. But I think industrywide, we lacked some engagement, we lacked some warmth, we lacked some relevance, particularly through digital channel. We need to re-instate humanization into digital, and AI is a tool that’s going to help us get there. Combined with in-person contact, AI can help deliver more personalized, more relevant, services in tune with what customers need at a particular moment.”

First Tech Federal sees greatly personalized interactions and services to customers as the main goal of its own expanding AI efforts. “Using AI and ML, we believe we will put ourselves in the best position to help each individual member meet their needs at any point in time,” says Upton. “We’re looking to be relevant when the member needs us to be relevant, no matter which touchpoint they so choose. We’re looking to leverage this into personalization and relevant engagement, whether it be a sales engagement, a servicing engagement, or a retention engagement.”

While AI promises to open up the levels of services financial institutions can deliver, there are challenges that need to be overcome, including skewed expectations, skills issues, and implementation issues. “Talent scarcity is a key gating factor,’ says Bjorn Austraat, senior vice president and head of AI acceleration at Truist. “This is true for specialized data science resources but – importantly – also for all-important resources and leaders that can speak to both technical and business stakeholders,’ he explains. This includes those “that are fluent in data science and executive-speak. An over-reliance on purely technical skills can lead to disjointed science experiments without a clear business return and an excessive focus on business outcomes — especially early on in sometimes lengthy data science and model ops lifecycles — can squelch disruptive innovation.”

Barriers to achieving success with AI are common across all sectors, says Charlene Coleman, senior managing partner and head of the modern finance sector of Launch Consulting Group. But financial services brings it ow sets of issues. “Deploying AI to democratize the financial system requires bold, human-centered leadership willing to invest in technology and talent. Next, institutions lacking an AI strategy will not move beyond the experimental phase. Most do not have a centralized data backbone that supports analysis and intelligent recommendations. Finally, they must adopt a new operating model that moves away from functional silos to enable speed and agility.”

Artificial intelligence “can help redefine and restore personalized experiences that build trust for consumers and small business owners,” says Coleman. “Assuming informed consent, an example is AI-powered personalized conversational interfaces and biometric profiles that have shown promise in helping vulnerable consumers avoid debt traps fueled by late fees and inflexible payment schedules.”

This means more than building models to support algorithms, no matter how well designed.. “People often assume ‘we just need a great model to solve our problem,’” says Austraat. “However, the model is only five percent of the solution. The integration, instrumentation, validation, ongoing monitoring and ultimately dollarization are the other 95%.” The key is to “think of the model as the race car engine,” he adds. “You’ll need a lot of other things to win the race: gas, shocks, tires, a pit crew, and a driver.”

The key to AI success in financial services is to sell or promote AI adoption to the business. “I use a simple phrase to accelerate this alignment: ‘Whose life is going to get better, by how much and how do we know that?’” says Austraat. “If you can truly answer that question, you have covered all the bases from framing, to deployment, to value proposition and value perception and realization, to political air cover. Explainability trumps model performance in financial services. In particularly sensitive areas such as credit underwriting, banks and other institutions must balance the desire to innovate and use cutting-edge AI with the reasonable regulatory expectations around explainability, robustness and fairness. The hottest solution doesn’t always win, especially if it’s too much of a black box.”

This requires a much more holistic view of AI, beyond the lab or data science team. “You can’t just let the data scientists do their thing,” Austraat says. “A holistic teaming approach centered around cross-functional pods is critical to engage legal, risk, data engineering, implementation engineering, operations, support and business leaders early and often to create sustainable success.”

In the end, technologies such as AI and ML “are simply tools,” says Upton. ‘You need to have a very clear business strategy, a very good go-to-market strategy, and a very good operational plan to leverage those tools to create the experiences and drive business value. People tend to get enamored with the tool or the tech, but they’re not really clear on their use case for the value of the investment. You can buy all the coolest tools in the world, but if you don’t account for the change management, the adoption, the helping the organization leading into the why and how to use these tools to drive out the things that matter, you’re just going to have a lot of expensive tools by themselves.”