AI Credit’s Next Test: Winning Customer Trust, Not Just Improving Underwriting
By CU Today Staff —
NEW YORK—The next competitive battle in artificial intelligence powered credit may not be fought over who can build the most sophisticated underwriting model, but over who can give consumers more control over when, how and why they use credit.
That is a central argument in a new report from PYMNTS Intelligence, produced in collaboration with Thredd, that examines AI-powered credit from the customer’s perspective. The report, Credit That Meets the Moment: What AI-Powered Credit Means for Customers, argues that AI, real-time data and tokenization are beginning to turn credit from a relatively static financial product into a service that can adapt to individual transactions and borrowers.
The authors contend the technology will matter little, however, if consumers do not understand or trust it. While AI can allow financial institutions to price risk more precisely and potentially extend credit to consumers who are difficult to evaluate using conventional scoring, customers see the technology differently: as a way to obtain credit when it is needed, with terms tied more closely to what they are buying and repayment options that fit their circumstances.
“This is the other half of the argument for credit powered by artificial intelligence (AI),” the report’s authors state. “Tokenization, real-time data and AI decisioning are not, from a customer’s perspective, tools for better risk management.” Instead, the authors say those technologies can make a credit line behave less like a rigid contract and more like a responsive service—one that can be instant, flexible and transparent.
One indication of that demand can be seen in buy now, pay later. Citing J.D. Power’s 2026 U.S. Buy Now Pay Later Satisfaction Study, PYMNTS Intelligence reports 52% of customers using a fixed-payment BNPL plan linked to a credit card decide to use the option after the purchase has occurred, compared with 48% who make the decision at checkout. Thirty-seven percent of U.S. consumers had used BNPL during the previous 90 days, five percentage points more than a year earlier. Adobe data cited in the report found BNPL accounted for $20 billion in online spending during November and December 2025, up 9.8% year over year.
But the potentially bigger change involves who can get credit and how that credit is priced.
The report points to Federal Reserve data showing roughly 32 million U.S. adults remained unscorable using conventional data—about seven million without a credit file and another 25 million with files too thin or stale to produce a reliable score. PYMNTS Intelligence argues real-time information about transactions and cash flow can provide lenders with a more complete picture than a static credit file.
Experian, for example, estimates its Lift Premium model, which incorporates rental and utility payments and cash-flow information along with traditional credit history, can produce scores for 96% of U.S. adults, versus 81% using conventional data alone. The report also cites a LexisNexis Risk Solutions survey of 875 lending and risk professionals in 10 countries in which 67% said alternative data increased their confidence in lending decisions, while roughly three-quarters reported improved portfolio performance.
Better Underwriting Part Of The Equation
For credit unions and other issuers, however, better underwriting is only part of the equation. Transparency may determine whether consumers actually embrace AI-driven credit.
PYMNTS Intelligence said consumers should be able to understand and control what their credit is doing on their behalf. It notes U.S. and U.K. regulators have reinforced requirements that borrowers receive clear, specific reasons for credit decisions regardless of the complexity of the underlying model. The report points to emerging products that let customers establish rules determining whether particular purchases draw from debit, installment credit or another credit line—and change those rules as circumstances change.
That capability could become especially important as younger consumers move into their prime borrowing years. PYMNTS Intelligence research cited in the report finds 40% of Millennials and 48% of Gen Z consumers would increase card usage if they could choose their repayment plan at the point of sale. Nearly 60% of consumers overall—and 70% of “bridge Millennials”—prefer rewards and payment structures that can adapt rather than remain fixed. Nearly seven in 10 cardholders also say the quality of a card’s app influences which card becomes their primary card, with the percentage even higher among younger consumers.
The implications extend well beyond adding AI to underwriting. The authors argue issuers that treat AI-powered credit merely as a back-office efficiency tool could miss what consumers increasingly expect from the technology: greater choice, speed, flexibility and control.
As the report puts it, “None of this is a rewards story. It’s a control story.”
Originally reported by CU Today.