Executives cautioned that the rollout of generative AI will likely result in missteps that will capture the attention of regulators. Market participants will need to tread carefully.
Many bank executives are preparing for the impact generative AI is likely to have on their businesses. At a conference last week, bankers shared three use cases where they see a lot of potential: writing code, personalizing wealth advice and having more organic, thoughtful chatbot conversations.
Amid strong interest among financial institutions in generative AI — 60% of large banks with greater than $500 billion in deposits made tangible investments in generative AI by July 2023, according to research from EY-Parthenon — most aren’t yet at the stage where they’re rolling out customer-facing offerings based on it. But as companies work on internal trials, the longer-term implications on operations — including efficiency gains — are becoming clearer, said executives speaking at the Future Digital Finance conference in New Orleans last week.
Filling skills gaps
Sacramento-based Golden 1 Credit Union is exploring the use of generative AI to support its tech organization.
“Various flavors of AI have already been quite beneficial,” said Jay Tkachuk, executive vice president and chief digital officer at Golden 1 Credit Union. “We use gen AI to write our core code.”
The credit union experienced challenges finding workers that knew the old code in a legacy system, so instead of paying consultants exorbitant fees, it took a bet on a generative AI solution that delivered results, according to Tkachuk. AI outputs, however, go through several layers of human checks, and generative AI is so far only being used for “small tweaks” instead of major features, he noted.
“It may move toward that as we gain proficiency and expertise in the space,” said Tkachuk.
Personalized advice
Generative AI’s ability to synthesize unstructured data, suggest solutions for wealth management advisors and generate new content could prove fruitful in efforts to improve client experiences, argued Sam Palmer, who was recently named general manager of JPMorgan’s Chase Sapphire credit card program and formerly head of product and experience at J.P. Morgan Wealth management.
“Imagine that we connected all of these three things, we use gen AI to create something that is really personalized — like a premium package for [wealth] advisors,” he said. “Imagine that we’re able to take all the information that we have about the client … the AI can pull relevant information based on clients’ assets and bring topics or conversations or suggest resources to send to the clients.”
The speed with which generative AI can offer relevant advice could be a game changer, he suggested.
“The beauty of this is with a click of a button, an advisor will be able to generate something like this and go into a meeting, truly making the client feel like we recognize their entire relationship with the bank,” he said.
Given that the rollout of generative AI will evolve and change, Palmer said companies may want to start with small use cases, including those that drive improved efficiency.
“If you have small cases, you can pivot, you can change,” he said. Companies might make mistakes along the way, so smaller-scale use cases will let them course correct, he added.
Powering internal employee and customer conversations
Michigan State Credit Union is exploring generative AI use cases, starting with the rollout of Microsoft Copilot for employees. Copilot is a generative AI-supported assistant that can help draft text, summarize information and analyze data.
“By doing it in an enterprise way, we feel that we have that confidence that it’s happening in a way that we approve of and there’s no more risk concern,” said Benjamin Maxim, chief innovation officer at Detroit-based Michigan State University Federal Credit Union.
He said generative AI will help employees find information more quickly. The credit union is looking at a use case around querying contracts through generative AI that’s been successfully rolled out elsewhere, he noted.
On the member-facing side, Maxim said the credit union is working on opportunities to offer generative AI-driven conversational capabilities, which it expects to roll out this summer. Instead of a ChatGPT-style dialogue, the credit union is looking at what it calls “quote mode,” a conversational format in which the generative AI tool will offer reference links where members can go back to original sources of information.
“Quote mode is really ‘Hey, here’s some information, we found it here. If you want to verify it, you can go over here,'” said Maxim. “There’s a lot of opportunity to democratize access to financial education through conversational AI … people don’t have to feel embarrassed that they don’t know the answer.”
The longer-term view
Executives cautioned that the rollout of generative AI will likely result in missteps that will capture the attention of regulators. Market participants will need to tread carefully.
“In general, the impact will be quite beneficial, but you can almost guarantee that spectacular mistakes will be made which will then prompt regulation and so there will be bumps along the way,” said Tkachuk.
Despite the risks, AI in the banking sector is being rolled out in a measured way — as “the right tool for the right problem” — instead of a blanket approach, said Rajneesh Vijh, senior vice president of global technology at Bank of America.
“The way it’s being applied right now in the banking sector is the appropriate way, which is, AI is task oriented,” he said. “We’re using AI to solve specific problems, for the benefit of our customers.”
