What Accountholders Want From AI Is Anything But Obvious
By Nicole Volpe, Contributor at The Financial Brand
Simple Subscribe
Subscribe Now!
Financial institution strategists looking to deploy AI are likely to overemphasize its technical and institutional implications in large part because they feel urgent. AI can affect an institution’s competitive position: If rivals move faster, will my institution fall behind? AI has the potential to improve profitability: Can AI materially improve our operations and financial performance? And AI forces a conversation about risk: What happens when AI gets something wrong?
As a result, understanding what the consumer wants from AI may sometimes get less attention — and may even be taken for granted — because those benefits can seem almost self-evident. Most bank and credit union leaders have used ChatGPT, Claude, or Gemini themselves; many of their institutions already operate first-generation customer service chatbots. It does not take deep analysis to picture AI making those experiences more relevant and useful.
But new research from MX suggests the accountholder benefit is anything but obvious. Consumers’ financial lives continue to become more fragmented and complex; they increasingly expect their financial institutions to know and understand them; and the everyday work of staying on top of their money has only gotten harder. Meanwhile, consumers remain highly selective about where they want AI to help — and where they do not want it to take over.
That combination uncovers an opportunity for banks and credit unions, according to Eli Hiller, Principal Product Manager, AI Strategy at MX. “AI that helps consumers understand, navigate, and improve their financial lives without asking them to surrender control.”
The survey, which draws on responses from more than 1,000 U.S. adults, offers a clearer picture of where consumers want AI to show up in their financial lives, and what banks and credit unions can do to earn permission to go further.
Want more insights like these? Check out MX’s content hub: Data in Action
AI Adoption Is No Longer the Hurdle. Trust Is.
Banks and credit unions do not need to persuade consumers that AI can be useful. Nearly half of consumers in MX’s survey — 49% — say they already use AI at least weekly, and 26% use it at least daily. They are using it for decidedly practical purposes: searching for information, writing and editing, learning, shopping research and personal organization. Younger consumers, particularly Gen Z and Millennials, are leading that adoption.
But familiarity with AI does not carry over automatically when money enters the picture. Only 32% of consumers say they trust AI to help manage their finances, while 50% actively distrust it. That gap between use and trust may be the most important starting point for financial institutions thinking about consumer-facing AI.
It also changes the nature of the adoption challenge. Banks and credit unions are not introducing consumers to an unfamiliar technology. Increasingly, they are introducing AI to people who already have expectations for what good AI feels like — and who may become more cautious, rather than less, when the stakes rise from drafting an email or researching a purchase to managing debt, savings, or investments.
The generational differences sharpen that point. Gen Z and Millennials are significantly more likely to use AI regularly, while Baby Boomers remain much less likely to do so. That means institutions cannot assume a single starting level of comfort across their customer base. But neither should they assume that frequent AI users are ready to turn over financial decisions to it.
For banks and credit unions, then, the question is becoming less whether consumers will use AI than what they will trust it to do.
Consumers Want AI as a Coach More Than a Decision-Maker
The trust gap comes into clearer focus when consumers are asked about specific AI use cases. What emerges is less a preference for advice over execution than a weighing of risks.
Consumers are relatively open to AI that delivers reminders, breaks down spending, categorizes transactions, provides personalized recommendations or offers basic financial advice. Such applications make a complicated financial life easier to manage while leaving the final decision with the customer. As many as 47% of consumers say they are ready to trust AI in at least some financial-guidance use cases.
That openness extends even to AI that acts on their behalf, but it depends on two factors. When stakes are low and actions are easily reversed, consumers show strong interest. But when stakes are higher, knowing the AI is controlled and governed by a financial institution they already trust matters most. There’s a meaningful difference between asking a general-purpose AI tool to apply for a loan and using a branded financial experience where you know your trusted institution is handling the request on your behalf.
MX tested agentic applications ranging from canceling unused subscriptions and optimizing credit-card usage to refinancing debt, applying for financial products and automatically switching a checking or savings account. Interest in the lower-stakes tasks runs about as high as it does for guidance — up to 47%.
But consumers withdraw trust when it comes to applications in which AI makes a costly, hard-to-reverse decision on its own — approving or denying a loan, or automatically moving money between accounts. Support for those falls to 32%.
“What makes the difference is less about AI itself than about balancing consequences and control,” said Hiller. “Consumers appear most comfortable when AI can surface an opportunity or recommend an action — but they can decide what happens next.”
That gives banks and credit unions a useful starting point. The strongest early consumer applications may be those that make financial decisions easier without taking those decisions away from the customer.
Financial Institutions’ Existing AI Experiences May Be Setting the Bar Too Low
If consumers are cautious about financial AI, part of the reason may be that what they have encountered so far has not given them much reason to expect more.
Only 34% of consumers in MX’s survey say they have used a general-purpose AI chatbot in financial services. More strikingly, 53% of those who have used one say the experience was not very helpful or not helpful at all.
That matters because AI tools from financial institutions are likely to shape consumers’ expectations for what financial AI can do. A general-purpose tool that struggles to answer a question, misunderstands what a customer wants or simply adds another layer between the customer and a solution creates a poor first impression. But financial AI built specifically for banking — AI that understands context, integrates with consumer-permissioned data, and actually solves problems — can demonstrate what’s truly possible
The lesson for banks and credit unions is not simply to build a better chatbot. It is to avoid treating the presence of AI as value in itself. Consumers already have access to increasingly capable general-purpose AI tools. A financial institution’s advantage lies in what those tools do not have: an understanding of the customer’s actual financial circumstances and the ability to use that understanding to solve a relevant problem.
MX’s findings suggest that usefulness may be the bridge to greater trust. “Consumers are willing to give AI a role when the benefit is tangible and the stakes are clear,” Hiller said. For financial institutions, the task is to make the first experiences good enough — and useful enough — to earn permission for the next ones.
Useful AI Depends on Actually Knowing the Customer
The promise of financial AI rests on something more basic than the technology itself: whether the institution has enough information to understand the customer it’s trying to help.
Consumers increasingly expect that understanding. Sixty-one percent of respondents say their financial provider should know them and understand their financial needs. Yet many still see evidence that institutions do not. More than a third say they often receive messages that are not personalized or relevant, while 31% say the insights their provider offers are often irrelevant or outdated.
AI raises the stakes around that gap. A generic marketing message is easy to ignore. An AI-generated recommendation based on an incomplete picture of a customer’s finances can be irrelevant at best — and potentially damaging to trust.
The challenge is that a customer’s financial life may extend well beyond what any one bank or credit union can see. MX’s research notes that users connecting accounts to its platform actively link an average of three accounts, while transaction data reveals another three “discovered accounts” on average.
“Even consumers themselves may not have a complete picture of how fragmented their financial lives have become,” Hiller said — so it’s essential for institutions offering AI applications to have connected, current financial data. Consider the use cases: recommending how much someone can safely save, identifying unnecessary spending, flagging an approaching cash-flow problem. An AI tool cannot credibly advise someone on saving, spending, debt, or cash flow without context, without having a view of the broader financial picture.
For banks and credit unions, this may be where the opportunity and the burden meet. Consumers want AI that understands their situation well enough to be useful, but they are reluctant to give it unchecked authority. The institutions that earn greater trust will therefore need to demonstrate both sides of the bargain: a sufficiently complete understanding of the customer, and restraint in how that understanding is used.
