- Key insight: In building their own AI tools, banks must keep humans embedded at high-stakes moments and give customers transparency and control over AI's involvement, according to executives from JD Power and PwC.
- Expert quote: "When it comes to customer service, I still believe that the differentiator is going to be human touch." —George Korizis, PwC front office strategy partner and transformation leader
- Forward look: The smart path forward doesn't involve chasing every AI use case at once, experts said. They also suggest measuring success by gauging whether customers actually use and like tools, not just whether they shipped.
Artificial intelligence is challenging banks' long-standing role as the first and most trusted stop for financial advice.
JD Power's
Still, banks haven't lost their footing altogether. Jim Miller, who serves as vice president of financial services at JD Power, stressed that banks still have a trust advantage over tech firms and fintechs when it comes to handling customers' financial data — but he said that advantage is a use-it-or-lose-it asset. If banks are not "proactively" building and promoting AI-powered tools that leverage this trust, Miller warned, customers will continue turning to generic AI tools or fintech competitors.
"AI in some ways has already caught up with the bank, as far as helping make smarter financial decisions," Miller said.
If customers are going to use a traditional financial institution for advice, those companies have to get over the "hurdle" of demonstrating to customers that they aren't just selling another product, Miller noted. But once they do that, they benefit from a "greater degree of trust" that the bank will "protect their data" from both business and regulatory perspectives, he said.
"If you go to an open-source LLM, it doesn't have the bias of trying to sell you something, but then it also doesn't have the obligation to protect your data or to give you good answers," Miller said. "If you have a model that's just as likely to pick up bad financial advice as good financial advice, that's a concern."
Consumers still want a human in the loop
While consumer engagement with AI tools has been "growing faster than anyone expected," consumer trust and comfort still hinge on human oversight at the decision point, emphasized Roberto Hernandez, front office strategy partner at PwC.
And though consumers are increasingly using AI for lending-related research, they still want a human involved in the actual lending decision, according to PwC financial services industry leader Peter Pollini. He cited a
PwC found that 91% of financial-services industry executives say technology and AI tools are becoming more important for helping clients navigate uncertainty — but only 53% of consumers trust AI-powered financial tools during times of market volatility.
Read more:
Should a bank ever be liable when a customer gets scammed ?OpenAI models' Hugging Face breach is a red flag for bankers The top-performing banks with $2B to $10B of assets in 2025 Inside Upstart's journey to get a bank charter approval
Another PwC survey — this one on mortgage and auto lending — found a clear divide in how people want AI to be used during different parts of the borrowing journey, according to Hernandez.
During consumers' research phase, they are happy to use self-service tools to compare options, the survey found. But when it comes time for the lender's decision, most consumers want a human involved — both to answer complex questions and to ensure that a person, not just an algorithm, reviewed their file.
"When it comes to customer service, I still believe that the differentiator is going to be human touch," said PwC front office strategy partner and transformation leader George Korizis. "It's just the level and the degree to which companies are going to still provide that, and when they're going to choose to do that."
He emphasized that consumers want two things from AI offerings: transparency and optionality. Clients want to know how AI is being used, and they want to have control over which parts of the banking experience are infused with the technology, according to Hernandez.
Younger consumers — and more stressed ones — are leading the AI shift
Miller highlighted a finding from the JD Power survey that shows younger consumers are relying more heavily on AI tools. The same demographic is gravitating toward large national banks and fintechs at a higher rate than other consumers, he added.
"We're seeing national banks and fintechs get the largest share — beyond what you would expect from them — of new accounts," Miller said. "It's really coming at the expense of regional, midsize and community banks."
While 27% of consumers told JD Power that they found AI somewhat or significantly helpful in managing their personal finances, that rate shot up to 40% for those younger than 40 years old.
Among "overextended" consumers — those who are financially vulnerable or stressed — the number rose to 48%. Miller attributed that group's increased reliance on AI to people's tendency to "blame themselves for their poor financial situation, whether it's their fault or not," which may make them reluctant to discuss their money problems with a human being.
"They're generally not the ones going into a bank wanting to talk to somebody about their financial condition. It's kind of like not wanting to go to the doctor and tell them you don't exercise," Miller explained. "The anonymity of AI would be very beneficial to those clients. They don't have the embarrassment."
Miller said the diverging demands of different client segments actually give banks an opening to deploy AI effectively, using it to customize offerings on a segment-by-segment basis rather than treating clients uniformly.
"Don't take your eye off the ball of serving today's customers, while you build an AI–agent enabled infrastructure for the future," Pollini said.
How banks can build AI tools that actually work
Hernandez's advice to his banking clients is to stop looking for a way to "inject AI at a particular step in the journey." He recommended that firms bifurcate their approach to AI innovation.
"Before you start talking about where AI might be transformational, or any type of automation or technology might be transformational, identify the steps in the process where your mission is just for the target customer to be able to complete that process in the most efficient way, and which are the segments of the process where you really want to deliver delight," Hernandez said.
Hernandez and Korizis laid out three metrics for the successful implementation of an AI tool: whether consumers use the tool, whether they like it, and whether the technology achieves the goal it's tasked with accomplishing.
Korizis has observed a common failure as companies roll out AI tools: A product starts with a single objective, but the number of goals balloons to four or five by the time it ships, muddying the firm's ability to measure success. Ultimately, he said, customer feedback will reveal what actually worked.
"What separates really leading-edge organizations from those that are the mean is the ability to adapt to that feedback and to actually show that it's being incorporated" when "they release the next features," Korizis said.
As banks deploy AI tools to their clients, Korizis observes them pushing "certain initiatives to get to parity on perception and brand" with third-party options. Banks are rolling out products to "showcase that they're evolving with the times," Korizis said, instead of approaching their agentic feature development as an opportunity to "build on trust."
"Don't break the trust that you have with your customers. That means don't be reckless, don't try to do things that introduce risk," Korizis advised. "Know your audience. Adapt to personalization, and be able to apply personalization that is truly catering to your customers' needs."
