Navigating the Landscape of Generative AI in Community Banking

As part of my final project at GSBC (Graduate School of Banking, Colorado), I was able to choose a strategic topic to work on with a group of forward-thinking bankers. The school calls these projects “Signals” as the nature of the research is all about future impacts on our industry.

I have been intrigued by the storm of information related to Generative AI (GenAI), so I decided to dive headfirst down this fascinating rabbit hole. When researching GenAI, I had several questions in mind that I wanted to be able to answer: 1. Is it real or just a fad? 2. How do we leverage this technology to generate financial benefits for our institutions? 3. What are the biggest Risks associated with using this technology? 4. How do we overcome these Risks, so we feel confident in implementing this technology? 5. How are regulators going to respond/catch up to the technology? If we can answer all these questions, I hope we can figure out what next steps we can take to benefit community banks with the technology.

The first question was pretty easy: nothing in the history of the Internet has grown as fast as Chat GTP and other AI models in terms of users. Something that can gain 100 million users in a matter of months cannot be ignored. A recent article in the Wall Street Journal had interesting quotes from both Elon Musk and Jamie Dimon reiterating the transformative potential of GenAI. Musk states “AI is the fastest advancing technology he has ever seen.” While Jamie Dimon stated in his annual shareholder letter that “AI could be as transformative as some of the greatest technological changes of the last several hundred years,” comparing AI to electricity and the printing press!

The next question related to how the technology can be used is unclear, just because there are so many applications! The challenge here will be narrowing our focus to applications that have big impacts and finding the right partners. Due to the cost involved, community banks are not going to be able to develop their own proprietary GenAI models. Thankfully, large and reputable vendors, such as Microsoft (CoPilot) and Google (Bard), quickly respond to the changing environment and roll out offerings with robust models backing the technology. Ceto was able to sit down with Lamont Black, a professor at DuPont University and expert in future technologies related to financial services, to discuss some of the practical uses of AI in financial institutions. He suggests that AI will become a great collaborator to help people become more efficient, while not replacing humans in most job functions.

Questions 3-5 are all related, what are risks, and how do we overcome them while keeping ourselves in the regulator's good graces? This is where the questions get harder. The first thing I want to point out is that we truly do not know all the risks associated with this technology, and this is likely the biggest risk: the unknown. While I was writing this, a novel GenAI Fraud was making world news, where a company was defrauded out of $25 million using Deep Fake technology.

In addition to fraud, Gen AI has several other serious issues that are known and must be overcome. Embedded bias is when the technology is trained on accidentally biased information and carried forward into decision-making. Also, there is Hallucination risk, where AI gives believable responses that are actually false. Explainability is a real challenge that must be overcome. If AI is making a decision, we must understand and be able to explain it to a customer or regulator. The regulatory agencies have not caught up with the technology yet, so we do not really know how they are going to react...but we know they will.  A presidential Executive Order in October of 2023 guarantees this response.

So, what are bankers to do? First, we must pay attention and develop a plan. This is likely the most transformative technology we have seen since the creation of the Internet. That sounds like hyperbole, but I believe it is accurate. Going back to question 2, how do we use it, and use it safely? I think partnering with a technology company will be necessary, it is simply too complex for anything that a Financial Institution can create. Even Chase and Wells Fargo are not trying to develop this on their own. Chase has stated they have over 2,000 employees working in AI-related development initiatives! Community Banks don’t have that scale.

While we cannot overcome the risks of the unknown, there is a way to limit our exposure to additional risks. Many of the large FIs are looking at how they can use this in a customer-facing manner. For community banks, leveraging the technology internally vs externally will greatly limit the risks associated with using the technology, while also improving efficiency. GenAI, as an internal support expert or first-pass underwriter, can greatly boost a community bank's efficiency. I often discuss with FIs the concerns they have with concentrations of organizational expertise. I have been told that if a certain person left the bank, they risk losing knowledge that could take years to replace. Being able to utilize GenAI to digitize this institutional knowledge would dramatically reduce this risk. Once digitized, this information can be queried by many people, freeing up that internal expert to do their day job! In many markets, the right people are not available to fill the roles we need; using GenAI to help the people we have become more productive can bridge that gap.

While pondering how to leverage this technology, we must also consider how other industries will use it. This has the potential to create additional opportunities, while also creating additional risks. The rapid adoption of the technology will initiate increased technology spending for many of our customers. Rolling out specialized financing programs targeting this technology could increase loan demand. However, we also need to be aware of whether we have a concentration in a certain industry or geography that is going to be heavily impacted. If I am a small FI in San Francisco and many of my customers are programmers at technology companies, what happens if these companies no longer need 90% of their programmers due to this technology? How does that impact my formerly affluent customer base? Energy usage is also a big concern; large AI data currently consume around 4% of US energy consumption...that could spike to 25% in as little as 5-6 years, according to Rene Haas, the CEO of ARM.

GenAI is a powerful new technology that must be a part of any FI's strategic planning and risk management. Banks must consider how to use it safely for their own benefit and the larger economic and specific customer impacts it will have in the near future. A McKinsey study has pegged the overall economic impact of GenAI at $2.6 - $4.4 TRILLION worldwide, something with that much impact must not be ignored. The technology has many additional uses, and we are just scratching the surface. However, part of using this technology responsibly is not forgetting to include human checks and balances to ensure we protect ourselves and our customers. The human element will be critical to overcoming issues of bias and explainability highlighted above and ensuring we are using the technology effectively. I’ll end with what Jamie Dimon said in his investor letter, “While we do not know the full effect or the precise rate at which AI will change our business—or how it will affect society at large—we are completely convinced the consequences will be extraordinary.”