Tech leaders say AI could end humanity. Banks still love it

• Key insight: As some tech insiders warn that rogue bots could wipe out humanity, banks' continued enthusiasm for AI has fallen out of step with public opinion.
• Supporting data: According to a recent CNN poll, 75% of American adults view AI with "fear and concern."
• Expert quote: "Every time in the history of the world that you've had this transformational technology, these types of fears come up. … But almost every single time, what they end up doing is being incredibly enhancing to humans and their capacity and productivity." —Brendan Coughlin, president of Citizens Financial Group

This month, anxiety about artificial intelligence skyrocketed as industry insiders warned that the technology could end humanity. But by all appearances, that hasn't put a dent in banks' enthusiasm.

"I am by no means a buyer of, 'There's a 10% risk that humans don't exist in a decade,'" Brendan Coughlin, president of Citizens Financial Group , told American Banker. "I think there is more good that these tools will do long-term than those kinds of catastrophic risks."

For years, banks have embraced AI both in their own operations and, amid the data-center construction boom, as a source of loan growth. Internally, banks have deployed AI everywhere from back offices to call centers, while externally they've loaned billions to companies involved in the building and powering of data centers.

But in recent weeks, that gung-ho attitude has collided with a wave of safety concerns about the technology. In July, a rogue OpenAI model hacked into the servers of the tech startup Hugging Face. In early September, a researcher at the AI firm Anthropic abruptly resigned, warning that AI "could kill us all by the end of the decade." Soon afterward, three AI moguls publicly echoed his concerns.

Since then, public opinion of AI — which was already on a downward trend — has plummeted. A recent CNN poll of U.S. adults found that 75% of respondents regarded AI's development with "fear and concern," while just 25% felt "hope and excitement."

At banks, however, hope and excitement still seem to be the prevailing attitude.

"AI will have an impact on work," Bank of America CEO Brian Moynihan said Thursday during a Punchbowl News interview. "It'll enable us to do more — more revenue, more profit, more activities, and take away work that, frankly, people would rather not do."

At Citizens, Coughlin has been leading a three-year, $300 million project called "Reimagine the Bank," which largely focuses on making operations more efficient through the use of AI. Have the recent fears of an AI apocalypse changed any of that? Coughlin's answer: "Not really."

"Every time in the history of the world that you've had this transformational technology, these types of fears come up," Coughlin said. "And you've got to be thoughtful and manage them well. But almost every single time, what they end up doing is being incredibly enhancing to humans and their capacity and productivity."

At Citizens, he said, this faith in new technology has already been rewarded. Thanks to AI, the average engineer at Citizens is now 30% to 40% more productive than they had been a year ago, according to Coughlin — and the future looks even brighter.

"As it scales even further, you can imagine a world where … the role of an engineer actually becomes a little bit more managerial," Coughlin said. "You're running eight to 10 agentic [AI] agents. … So our engineers become our more senior engineers, and they're directing the agents."

Read more:

• AI might be in its bubble phase. How exposed are banks?
• 11 ways banks can counter the threat of AI agent swarms
• Overheated data center knocked 23 credit unions offline
• How banks are looking to tame their growing AI bills

Of course, banks generally don't use the kinds of frontier agentic AI models that have launched cyberattacks. In fact, some lenders have said they plan to deliberately use older, less cutting-edge models for certain tasks in order to save money.

"One of the things that we and others are going to have to optimize over time is the cost dynamic there," Coughlin said. "Depending on the task for an engineer, it may make sense to use a model that is cents on the dollar versus a frontier model for a really sophisticated task that is super-expensive."

Bank executives are not oblivious to the safety concerns. Both Moynihan and Mark Wiedman, president of PNC Financial Services Group , directly addressed those concerns at a conference in New York this month, just days after tech leaders warned of the potential for AI agents to break free of human control.

"This weekend was a lot about the risk in AI, and that largely is around agents just left to operate," Moynihan said. "We just don't do that."

For his part, Wiedman said PNC takes the danger of rogue AI agents seriously.

"Making sure agents are doing what they're supposed to do and not doing something else is a top priority for us," he said. "So we've got to walk cautiously because that's going to be a challenge, I think, for every large organization using AI."

But he quickly switched gears to the technology's more positive potential.

"Is the agent going to be a little bit too aggressive in what they're trying to get done? We're trying to make sure we keep that under control," Wiedman said. "But broadly, it's a big opportunity."