GenAI is part of many conversations these days. And it’s true the technology has the potential to transform businesses — improving customer support, risk management and even commerce itself. But as Sam Hamilton, head of AI and Data at Visa, told Karen Webster in the latest installment of the “What’s Next in Payments: Memo to GenAI Companies” series, the transformative aspects of generative artificial intelligence (GenAI) can only be realized with the right inputs fed into the LLM models, at the right time.
Visa sees more than 300 billion transactions move across its network annually, tied to $15 trillion, and as such has the right data to feed into those models to support new innovations in payments and beyond, while fine-tuning its own internal capabilities. Through the past five years alone, the firm has invested $11 billion (and counting) in cutting-edge technology, said Hamilton. In one example, through the use of AI, Visa blocked roughly 85% more suspected fraud last year on Cyber Monday last year compared to 2023.
At a high level, said Hamilton, “I can’t think of one area that GenAI is not going to transform,” building on the foundation of trust that’s already been built between the customer and Visa. “There’s going to be dramatic change … the AI timeline is very fast,” he said. The large language models (LLMs) are “progressing nicely” according to Hamilton, so much so that the ways in which we interact with computers are changing, evolving toward a state where it may feel not like we’re talking to another human being, not a machine.
“We were the first network to deploy AI-based technology for risk and fraud management,” said Hamilton, who added that the company has been using artificial intelligence (AI) for more than three decades. At present, Visa’s technology platform provides more than 100 products with AI, the byproduct of $3 billion invested during the decade in which Hamilton has been with the firm.
“We’re getting ready for AI to be deployed across the company — and enable capabilities that power the next wave of payment security.” To get there, he said, data is critical, but so is technology — the hardware and software that power the GPUs and LLMs, developed by the right data scientists and other talent that help steer true innovation.
In recent months, the company updated its Visa Account Attack Intelligence offering, which generates risk scores for card-not-present transactions, identifying issuer bank identification numbers (BINs) and merchants that are under attack by fraudsters, who leverage AI to automatically test payment fields (in what’s known as account enumeration). Visa itself is attacked constantly — its APIs are besieged by fraudsters billions of times each month, so the company is proving especially adept at using AI to fight fire with fire.
“You need AI to do all this,” said Hamilton, who noted, too, that Visa has been working, through its $100 million GenAI fund to support startups and help develop frameworks to speed the development of new products and services.
GenAI is proving especially useful in client support — moving chat bots toward natural language interactions that converse directly with users, helping the payment network’s clients find the support they need from Visa. The Visa analytics platform, said Hamilton, also is used by corporate clients to find the insights they need about their businesses based on the data on hand.
“We’re also working hard to protect our ecosystem from risk and will use GenAI” for those purposes, said Hamilton. Key among those efforts is making sure that Visa credentials are secure and that the company’s systems are available at all times.
Beyond the “behind the scenes” uses of GenAI as described above, Visa is also harnessing its potential to change the shopping experience itself, giving rise to ever-more-personalized recommendations as AI agents “learn” about individuals and their preferences. Payments data can help Visa’s merchant partners and consumers deliver that robust experience — powered by consumer-permissioned information. Data tokens, he said, protect sensitive details about consumers.
“You’re in control of the data 100% of the time,” said Hamilton of consumers. “With the Visa credentials, you choose to share them with people when you need them and when you need them — and with the merchant, you can be comfortable sharing that credential to get the payment done … we’re seeing that this is very valuable.”
As he told Webster, “AI is no longer just a tool. It’s a collaborator and a partner in shaping our future at Visa.”
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.