For the past year, generative AI has dominated discussions about how emerging technology stands to transform our lives, and the payments space has been a big part of the conversation. Though generative AI is a hot topic, the road to development is long. Along with the opportunities come notes of caution and warnings that this revolution may take a while.
A report from Javelin Strategy & Research, Generative AI Comes to Life: Notes from the Field, takes a deep look at how companies in the payments space are making use of this capability. One of the conclusions the authors come to: Despite all the hype, don’t expect to see significant changes resulting from AI anytime soon. For one thing, in-house development of large language models requires enormously robust data to feed to AI. Organizations aren’t ready to fully capitalize on this yet, so the impact within the payments industry is further off than many people assume.
As much as AI stands poised to alter the way payments processors do business, the changes will be incremental.
“In the short run, we’re going to see simpler, smaller real use cases using AI,” said Christopher Miller, Lead Analyst for Emerging Payments at Javelin Strategy & Research and a co-author of the report. “But we still need to develop the whole infrastructure around it, and a whole understanding around the technology itself, so there’s not going to be an overnight change. Multiple years, I think, are required when we will start to be able to automate even more things than we can today, in terms of ingesting lots of information and understanding how to evaluate that information, taking into account your specific account balances or financial needs or preferences.”
AI has the potential to help a bank improve its back-office efficiency or reduce the time needed to transfer funds or decrease instances of fraud. That’s likely to be impactful to business results and might result in lower prices for customers, but it’s not likely to be very visible to outside observers.
Invisible Changes
One area where AI has already made changes is in client interaction. There are customer service actions that can provide suggestions to those customers based on their own history as well as the history of similar customers, but those, too, are instances that will likely be invisible to the users.
“Institutions are not going to expose it directly to customers,” Miller said. “But they will expose it to customer service reps who are going to use these tools to effectively be more productive in their interactions with people.”
One upshot for consumers might be that service calls become shorter because the representative is able to give the caller the answer more quickly. That might be almost unnoticeable for the caller, but for the business, it can make a big difference. Shaving eight seconds off a customer call might not affect the caller at all, but for organization (for example, a top-20 bank) that handles hundreds of calls every day, the time savings add up quickly.
Tools like generative AI could also lead to some customer segments getting better advice or guidance. This could also affect many parts of their financial life, such as assistance with investments or wealth management.
Payment processing is one area that could see big changes. AI is likely to offer more suggestions when a consumer is making a payment, because it knows all of the payment methods available to that customer. “In the moment of a sale, it could calculate that it’s best that you use your cashback card instead of your Costco card,” Miller said. “It can manage all those options for you and lock them in as a way of maximizing the transaction for both you and the merchant.”
Personal Communications
Customized communications are another example of something AI could improve for financial organizations, even if the customers never notice the change. There are thousands of reasons a company might want to communicate something to its customers, ranging from being declined for an account to acknowledging an address change to encouraging the opening of a new, different account. Those communications have to be vetted, approved, and made compliant with various regulations.
“When you get any type of communication from your bank, even notifications within an app, those are generally preapproved text,” Miller said. “The systems are limited in terms of how specific the communications about anyone can be, so they usually have a form approved that is ready to be sent to you. Imagine if you are able to, for example, have a system that has been trained on all of the relevant regulatory requirements for given areas and it can produce on the fly a letter that is both compliant and personalized. There’s some belief that this type of communication could be transformative in terms of presenting new opportunities to you or deepening the engagement that you have with the institution.”
It is likely to be years before we see the implementation of such transformative experiences as negotiations between individualized agents for unique payment terms. These require substantial infrastructural work by every member of the payments ecosystem to come to life at scale. But for organizations that eventually want to improve their processes via AI, the time to act is now.
