Artificial intelligence (AI) has left the pages of science fiction behind and entered the workforce.
And its latest evolution, generative AI, is already having an exciting impact on payments, helping to make transactions more efficient, secure and customer friendly.
“From a consumer perspective, I think some stuff is going to get way better,” Brady Harrison, director of customer analytics solution delivery at Kount, an Equifax company, told PYMNTS as part of the series “What’s Next in Payments: The Role in Artificial Intelligence in Payments.”
While the term AI is often used interchangeably with machine learning (ML), there has been a recent and significant shift toward true AI, especially with the rise of transformer models and neural networks.
In just one example of the benefits this brings, Harrison explained that while, historically, customers had to inform their banks of their travel plans to avoid card blocks when making international purchases, as card authorization issuing rules have become more sophisticated thanks to AI, the need for those types of notifications have basically gone away.
“Where historically you’d need to layer in some statistical analysis and machine learning, now we can use true AI where we’re looking at unsupervised, large anomalous data sets and making risks decision,” he said.
However, Harrison also pointed out that there are still opportunities for further improvement.
“There is always additional opportunity — and I think one hot area of focus right now is how to juice authorization rates,” he added.
Increasing the number of authorizations by working with issuers is particularly important for businesses looking to grow their customer base or retain existing customers.
“With the economic overhang, organizations are somewhat maxed out on new customer acquisition and are increasingly looking to grow and retain their existing customer wallet. We’re seeing machine learning and AI play a pretty substantial part of that on the risk side,” Harrison said.
Beyond allowing businesses to demonstrate their due diligence to issuers and ensure that authorizations are granted efficiently, AI and ML can help identify the right offers and incentives to drive further stickiness among customers.
Another area where the innovation is expected to have a significant impact is in instant payments, such as peer-to-peer transactions and customer-to-customer transactions, where AI is already opening new doors.
These real-time payment methods require robust fraud detection, as scams like authorized payee fraud become more prevalent. AI, particularly machine learning, plays a pivotal role in making instant payments more secure for users by monitoring for potential fraud and analyzing the entire network of activity, rather than just individual transactions.
“Machine learning is going to play a critical role in real-time payments,” Harrison said.
The shift from traditional statistical models to more advanced AI has had a subsequently profound impact on data analysis — and the things that organizations are able to do with that analysis.
With the advent of transformer models and unsupervised learning, companies can now create more refined customer cohorts and target them effectively.
As Harrison noted, instead of focusing solely on acquiring new customers, businesses are now leveraging AI to retain and grow the lifetime value of existing customers, driving long-term growth.
“We are past the heyday of grow, grow, grow in the 2010s, and I think most people are pretty saturated — now, people are really looking at how do I segment and find my super users, then retain them,” he explained. “And so, for merchants and sellers, it’s how do we keep people in our system by giving them relevant topical material… giving people the right offers contextually at the right time is key.”
Still, the use of AI in payments raises ethical and compliance concerns. Harrison highlighted the importance of striking a balance between utilizing AI’s capabilities and maintaining privacy and security.
While AI can offer advanced insights, it’s crucial to protect customer data, prevent unethical practices and avoid unintentional bias in decision-making.
“There’s also the element of the uncanny valley,” Harrison explained. “People don’t like it when you’re too predictive — and you don’t want to freak your customers out.”
Looking ahead, Harrison explained that the concept of Generative Adversarial Networks (GANs) is emerging as a cutting-edge innovation in the payment industry.
GANs have the potential to scale the identification of fraudulent activities by creating adversaries that generate increasingly sophisticated fraudulent behaviors. This technology allows for the development of AI models that can adapt and learn from the evolving tactics of fraudsters.
Near-future innovations like GANs show that the future of AI in payments promises to be an exciting one. As the industry adapts to new technologies and data analysis methods, it is poised to offer better services, greater security and more personalized experiences to customers.
