The AI conundrum: When it comes to latest technology, should banks embrace it, avoid it or something in-between?

Ask bankers how an artificial intelligence language-processing model such as OpenAI’s ChatGPT will impact banking, and it’s early enough that some admit that’s clear as mud.

Ask ChatGPT itself, and the AI-based chatbot has a lot more certainty about its utility to bankers … and it’ll even list out at least seven ways it’ll be of service.

Kevin Runyon. (Peapack-Gladstone Bank)

Of course, that’s based on the assumption that banks decide to use the technology. Not all banks are hot on it right now, according to Kevin Runyon, executive vice president, chief information officer and chief digital officer at Peapack-Gladstone Bank.

“Many banks have chosen to ban it from their employees using it,” he said. “For us, we’ve embraced it because we feel that it’s not going to go away and that, rather than putting our heads in the sand, we want to put policies and processes in place for how it’s going to be used.”

Runyon said there’s an untapped potential for bankers using ChatGPT for research purposes. It can save hours, he said, in getting more details on subjects and looking up banking policy and procedures.

How can ChatGPT help?

Want to know the argument the robot overlords themselves make? Of the many ways the ChatGPT platform suggested it could be used by bankers in the future, here are a few, based on a prompt ROI-NJ fed into its system:

  • Customer support: ChatGPT can serve as a virtual assistant for addressing customer queries and providing real-time support. It can handle routine inquiries, guide customers through basic transactions and offer information about banking products and services. This helps alleviate the burden on human customer support agents, providing faster response times and 24/7 availability.
  • Onboarding and account management: ChatGPT can assist customers in the onboarding process, helping them open new accounts, complete necessary forms and fulfill regulatory requirements. It can also provide support for account management tasks such as balance inquiries, fund transfers and account updates.
  • Product recommendations: By analyzing customer data and preferences, ChatGPT can offer personalized product recommendations. Based on a customer’s financial profile, transaction history and stated preferences, the chatbot can suggest suitable banking products like credit cards, loans or investment options.
  • Fraud detection and security: ChatGPT can assist in identifying and preventing fraudulent activities. By analyzing transaction patterns, account behavior and customer interactions, it can help identify potential risks and alert customers or relevant security teams to take appropriate action.

While he believes in the use cases, he also acknowledges that there’s valid concerns in the industry about the use of these tools from a privacy and regulatory perspective.

“The issue, which we believe we can manage through policy and training, is all about data security,” he said. “If you take client information to do an analysis (with ChatGPT), where that information goes and how it’s secured on the back end is unknown at this point. Banks feel as though they can’t control what employees will be putting into this technology.”

Bank leaders tasked with overseeing the cyber back end of their business, such as Timothy Green, a senior vice president and chief cyber defense officer for Kearny Bank, look at any technology’s potential (and, really, inevitable) weaponization. That includes everything from basic email services all the way through platforms such as ChatGPT, Green said.

There’s a whack-a-mole game all banks have to play with security vulnerabilities that come with utilization of new tools.

Timothy Green. (LinkedIn)

“With technology, here’s what is constant: You can read every article from every different news source and you’ll find cyberattacks are consistently increasing and getting more sophisticated,” Green said. “As the threats continue to advance, you need good threat intelligence to understand what’s happening and how it impacts you directly and, potentially, indirectly.”

There’s also the possibility for the cybercriminals themselves to find their own ways to utilize a platform such as ChatGPT.

“With ChatGPT, they’re trying to create some lefts and rights,” Green said. “If you try to create a phishing campaign, they’ve created some guardrails that identify when you’re trying to do something malicious and it’ll tell you it’s not designed to do that. Time will tell if ChatGPT can create or alter code for malicious intent.”

Right now, due to the perceived privacy issues, Kearny Bank remains on the side of not green-lighting the use of ChatGPT for its operations.

Whether that’ll change at the bank or others still on the fence remains to be seen.

“We’ve embraced certain elements of AI, including machine learning and natural language processing; we’re looking to see how and where generative AI like ChatGPT can benefit us,” Green said. “We don’t want to dive into a pool without understanding the depth.”