Headlines about the uptick in AI adoption in financial services have dominated 2019. Large national banks have already made significant strides with AI; Bank of America, Wells Fargo and JPMorgan Chase are just a few that are attracting notice with their AI offerings.
But because the news tends to focus on the bigger players, there’s a common misconception that implementing AI is a feat that only the largest institutions — those with vast resources and billions to spend — can achieve. However, this is a myth that must be debunked.
The fact is that community and regional banks can also use AI to lower overhead, improve quality and enhance the customer experience. Understanding how AI can help achieve those goals will require that these institutions have working knowledge of what AI is. When many hear “AI,” they think of it in its most advanced form; e.g. machine learning or deep learning, but that’s far from the only applications. Using AI at the community or regional bank level will require comprehending which type can be deployed to best meet the needs at hand. The descriptions below detail which how different tools can be applied to leverage their full potential.
RPA, or robotic process automation
Various time-consuming tasks that bankers do every day can be handled by machines. When determining which areas to automate first, banks should look for repetitive, manual processes (such as auditing invoices, tracking contracts or imputing data).
RPA will improve productivity, speed and quality, freeing the humans currently trapped in these roles to focus on other areas of the institution where they do jobs that cannot be done by machines. Given that the key differentiator for community and regional banks is their ability to deliver personalized services, freeing up these personnel presents the opportunity to innovate and execute initiatives that support this distinction.
Further, RPA is a particularly well-suited tool to prove the benefit of AI. It is not costly to deploy these robots, and ROIs are typically short term; i.e., weeks and months rather than months and years. Back offices within many community and regional banks still rely primarily on manual labor, and once RPA is proven in one use case, there will be others that naturally lend themselves to being improved in a way that impacts the bottom line.
Machine learning introduces another level of capability when compared with RPA. This tool is applied to complex tasks and learns where those tasks can be improved, going beyond repetitive processes (e.g., learning to recognize and evaluate the accuracy of complex invoices that have very different types of information and layouts).
As you might imagine, the more complex a type of AI is, the more it costs and the longer the ROI can be. However, the savings potential is larger, as is the tool’s impact on productivity and quality. To deploy machine learning successfully requires a bank to consider where a tool that is built to learn how to solve complex problems within certain code parameters will change for the better the way some processes are handled within the operations.
Specifically, look for processes that involve variations and complexity. Underwriting loans using machine learning to expedite decisioning not only improves productivity and accuracy within a bank, but it also delivers a “wow” moment to the customer. Invoice auditing, while not customer facing, is another ideal application for machine learning; invoices can be screened for overcharges or other anomalies before they are paid, with 97% accuracy. Because the machine can handle multiple types of factors, the program can review any invoice.
Deep learning is an artificial intelligence function that imitates the workings of the human brain in processing data and creating patterns for use in decision making. Deep learning uses networks capable of learning unsupervised from data that is unstructured or unlabeled. Many bankers find it untenable to unleash a machine that has this level of ability within a financial institution’s network, with the autonomy to find and solve any problem it finds anywhere within that environment.
However, this wide-ranging scenario is not typically how deep learning is used. For example, deep learning can be trained to not just look for patterns within a payment transaction, but also know when a pattern indicates fraudulent activity. When the machine detects this condition, it can alert an analyst to freeze the related account(s) and investigate the activity further. The takeaway about machine learning is that it can build on layers of data — sender, user, social media event, credit score, IP address, and a host of other features — that may take years to connect together if processed by a human being.
With that type of capability and speed, get used to the idea of using this tool sometime down the road when the business case demonstrates the impact could mean millions to the bottom line of an institution.
AI will continue to gain traction in the coming months and years, and it’s paramount that banks implement it to boost efficiencies and become better-run organizations. While there are many advanced forms of AI that banks need to begin considering, for now, community and regional banks that just focus on the low-hanging fruit of AI, i.e. RPA, will find the results are material. Implementing automation to improve back-office procedures will enable banks to significantly improve processes and ultimately better serve their customers. The machines are not coming — they are here.
Michael Carter is executive vice president at Strategic Resource Management (SRM), a Memphis, Tennessee-based management consulting firm that services financial institutions.
Want more BAI Banking Strategies? Sign up for our free newsletter!