How AI Is Transforming Payments

With advances in artificial intelligence, developing more and better payments apps efficiently is no longer a pipe dream.

Whether they’re automating repetitive tasks or predicting market trends and consumer behavior, payments companies are embracing artificial intelligence.

A key driver, especially in the back office, is the technology’s ability to automate repetitive tasks, improve fraud detection, optimize payment processing, and gather insights from large data sets—all while reducing the risk of human error, especially for data entry.

Despite those advantages, businesses tend to have a love-hate relationship with AI. At first, it is not uncommon to treat the technology skeptically because of concerns it will eliminate jobs and have a potential bias in algorithms used to analyze data that could lead to unfair outcomes. Then there’s the cost of deploying and maintaining the technology.

Over time, observers say, that skepticism fades, prompting businesses to cautiously deploy the technology in the back office. Eventually, users come to accept it and look for new uses outside the back office.

The payments industry is no exception. Several key players have initially deployed the technology behind the scenes, such as for fraud detection on the network and to evaluate weaknesses in financial institutions’ fraud-detection systems.

In this regard, Mastercard Inc. is using generative AI, which creates content based on large data sets, to generate synthetic transaction data that mimic fraud and can then be used to hone fraud-detection models.

More broadly, 41% of financial institutions are implementing AI in fraud detection and case management, which reflects a “fundamental shift” in how payments are processed, secured, and optimized, even if most of that is not seen by end users, says Gilles Ubaghs, strategic advisor in commercial banking and payments for the advisory service Datos Insights.

Winter Haven, Fla.-based SouthState Bank, for example, has incorporated AI into a chatbot that has reduced document search times from seven minutes to 32 seconds, according to Ubaghs.

Other back-office uses include customer authentication and the ability to automatically schedule, trigger, execute, and monitor workflows across a platform.

“While consumers may focus on front-end interactions, the data shows that back-office and mid-office capabilities are the initial focus for AI deployments,” Ubaghs says. “This suggests that the most impactful changes might be occurring in areas consumers never directly see, but significantly benefit from through improved security, efficiency, and reliability.”

Operating Efficiencies

What makes AI an attractive tool for fighting fraud is its ability to analyze patterns across millions of transactions in real time and spot anomalies that a human analyst would overlook.

In 2023, Mastercard says, it harnessed AI to protect more than 125 billion payment transactions for banks, financial institutions, governments and consumers. In the first six months of the year Mastercard declined more than 100 million fraudulent transactions before they could impact the cardholder, the merchant, and the card issuer.

“AI’s predictive analytics can foresee potential issues and optimize processes in real time, which is invaluable for keeping payments secure and efficient,” says Greg Ulrich, Mastercard’s chief AI and data officer. “Plus, AI’s ability to learn and adapt over time means it keeps getting better and better, making it a reliable and scalable solution for the payments industry.”

AI also helps streamline payment processes by automating such tasks as chargeback management and transaction routing, Ulrich adds.

Small and mid-size businesses can also benefit from AI, advocates say. Use cases include automated invoice recognition and expense categorization, according to BILL, a financial-operations platform for small and midsize businesses. A recent survey by the company reveals that 85% of small and mid-size businesses want AI.

“AI adds value in these particular use cases by helping businesses maximize use of their data to make real-time decisions, increase efficiency, and reduce risks of fraud,” says Mary Kay Bowman, executive vice president and general manager of payments and financial services for BILL. “When we think about AI, we need to focus on practical innovations that can simplify day-to-day activities right away without a big investment.”

One promising form of AI is Generative AI, which learns how to create new content based on data the system has been trained on. Generative AI is supposed to get smarter as new information is provided and as human handlers refine the output.

Business sectors McKinsey feels could see the most financial benefit from generative AI include banking, where the technology could deliver value equal to an additional $200 billion to $340 billion annually if all use cases were fully implemented.

AI’s operating efficiencies are expected to create a robust market for the technology. A recent study by management consulting firm Bain & Co. predicts the market for AI-related hardware and software will grow 40% to 55% annually and reach between $780 billion and $990 billion by 2027. While fluctuations in supply and demand are to be expected, Bain predicts the technology’s upward trajectory is here to stay.

In banking alone, the market is expected to grow from $19.9 billion in 2023 to $34.6 billion this year, according to Precedence Research, a 74% leap (chart).

Growing in Comfort

With AI having proven itself in the back office and all signs pointing to increased adoption, payments companies are looking to expand AI into customer-facing applications beyond customer-service chatbots.

Potential use cases include account-to-account bank transfers, peer-to-peer and bill payments for consumers, and business-to-business and business-to-consumer payments for businesses.

“AI in banking has been used for call trees and virtual servicing for a while now, but one of the next-level applications is using AI to take orders for more types of payments transactions,” says Jeff Bucher, senior product strategy manager for Alkami Technology Inc.

Initiating payments verbally or using predetermined models based on incoming bills, invoices, and the need to move money to others is the logical next step for AI in customer-facing apps at payments companies, Bucher adds.

AI can also be leveraged to provide a more personalized customer experience, such as suggesting specific payment actions based on that individual’s history. “With multiple payment channels and unique features for many of them, AI-based decisioning can support in picking the most efficient and least expensive method for each payment request,” Bucher adds.

On the B2B side, AI can develop better customer experiences through auto matching businesses and suppliers, auto populating invoices, syncing accounting systems with bill payment platforms, detecting potential fraud, and providing payment or funding choices for customers, Bowman adds.

BILL recently introduced Sync Assist, an AI-powered feature that enables small and mid-sized businesses to automatically synchronize their financial data between the BILL platform and their accounting software. The synchronization allows businesses to transfer bills, payments, and other financial information between the two systems.

“We are applying AI to more use cases to further simplify and personalize the [small-business] customer experience on our platform,” Bowman says.

As consumers become more comfortable interacting with AI-based apps, payments providers are expected to accelerate their adoption of the technology. About 62% of mid-size-to-large companies are expected to be using generative AI for banking and payments within the next 12 months, according to Datos Insights’ Ubaghs.

In addition, 50% of banks that are in the early stages of adopting AI report minimal impact from the technology on their business. Within the next three years, however, 90% of banks that have deployed AI expect to see at least moderate impact on cash management, according to Datos.

“That’s a significant indicator of growing acceptance and expectation in understanding AI’s potential, even as organizations work to fully grasp its capabilities,” Ubaghs adds.

Consumers overall may be growing in comfort with the technology, but that level of comfort varies by generation. “Younger consumers and businesses tend, on the whole, to be more open to this techology,” says Ubaghs.

‘Learning on the Job’

There is also a learning curve for companies. Some are learning how best to utilize AI on the go before refining its use cases, says Mastercard’s Ulrich.

“Success hinges on proving not just technical efficacy, but commercial viability,” he says. “Overcoming this requires investment in AI education and training for employees, as well as collaboration with AI experts and technology providers.”

Meanwhile, payment companies can expect to make mistakes. “Everyone is learning on the job with AI, including those in digital banking,” says Alkami’s Bucher. “The rise in capabilities and applications for AI means that more is possible and there is a race happening to roll out as many use cases as possible. This will inevitably bring mistakes and challenges, but over time an equilibrium will be achieved with experience.”

That equilibrium will require a clear strategy for rolling out the technology, as well as employee education and training and collaboration with AI experts and technology providers to ensure the right infrastructure and governance are in place to support AI initiatives, according to Ulrich.

“Fostering a culture of innovation and experimentation can help organizations adapt to, and fully leverage, AI technologies,” he says.