Long an obsession of science fiction writers, “artificial intelligence” in the modern era of fast-paced technological innovation is a term that is as ubiquitous as it is nebulous. For the payments technology industry, however, the term describes advanced analytical technology that has an outsized potential to improve the payments ecosystem for banks, payments processors, merchants and consumers.
In fact, financial services companies will spend US$11 billion on AI in 2020, according to an analysis by IDC — more than any other industry cited.
They’ll stand to make a nice return on their investment as well, according to PwC estimates. In North America alone, AI is projected to increase the GDP of the financial and professional services industry as much as 10 percent by 2030, driven by increases in both productivity and consumption.
No industry realizes the impact of AI more than payments. Payments technology companies were the most likely of all the banking sectors surveyed, to be using AI technologies in their operations, Consultancy.uk reported in 2017. More than eight in 10 payments divisions — 84 percent — reported using AI in 2017. That was nearly 20 points higher the next most popular sector, IT, and 44 percent higher than the third most popular, finance and accounting.
The potential for growth in this sector — from payments companies and tech companies to banks, retailers, etc. — is staggering. The AI market worldwide is set to achieve year-over-year growth exceeding 150 percent through this year, and it will continue to grow, with projections forecasting 127 percent year-over-year by 2025.
With such high expectations for the technology in payments and in myriad other industries, what will AI actually mean for consumers and merchants? What types of problems do payments technology companies and financial institutions expect to solve by deploying AI and other similar technologies?
Fighting Fraud Intelligently
The eternal struggle of the payments industry is to protect and secure the ecosystem from criminals bent on causing financial harm through theft and fraudulent charges. No industry is as committed to the fight than payments companies. They’ll invest more in advanced fraud detection and prevention technologies in the coming years than any other industry, according to Juniper Research.
After all, consumers are never liable if they’re made the victim of fraud through their electronic payments, so it is in our industry’s best interests to be aggressive and vigilant in fighting fraud in order to reduce losses.
We’ve done just that. Following the rollout of EMV cards — the “chip card” as it’s commonly known — counterfeit card fraud at brick-and-mortar retailers has declined by 80 percent, according to Visa. The volume of overall card fraud on card-based payments worldwide also declined in 2018 over 2017.
Still, as our retail experience moves toward e-commerce — environments where a physical card is not necessarily present — fighting fraud becomes more complicated. The trend among fraudsters is clear; card-not-present fraud grew 41 percent year-over-year between 2015 and 2018, increasing from 27 percent of cases in 2015 to 76 percent of cases in 2018. By comparison, card-present fraud accounted for only 19 percent of cases in 2018, down from 70 percent in 2015.
AI and machine learning are quickly becoming a valuable tool for payments companies and financial institutions to reduce fraud in all environments, but particularly in securing e-commerce transactions.
Through machine learning algorithms, payments companies can analyze more data in new and innovative ways to identify fraudulent activity. Every consumer transaction includes a lot of data, and with AI and machine learning, payments companies can search rapidly and efficiently through this data beyond the standard set of factors like time, velocity and amount.
AI, for example, can start considering a comprehensive mesh of multifactored logistical regression to create new dynamic weights for each data point when considering a transaction.
Most critically, systems can learn from each transaction, constantly improving and becoming more effective — something unique to machine learning and AI. In short, the use of AI can allow payments companies to look at transaction data in new and more effective ways, growing the amount of successful legitimate transactions while shrinking the number illegitimate ones that make it through.
With card-not-present fraud posing a $130 billion threat, according to Juniper Research, sophisticated fraud prevention and detection technologies powered by AI will be a growing area of investment for payments companies.
Further, AI and machine learning technologies have elegant applications to the world of underwriting — what I like to refer to as the “broccoli” of payments technology. If mobile payments, cashierless checkout and value-added solutions are “chocolate cake” for any payments tech insider, then they can be enjoyed only with a healthy dose of underwriting.
AI can help payments companies underwrite the merchants they onboard by creating a standard that is constantly adapting, living in real-time, and adjusting with every chargeback and instance of fraud. This not only creates a safer ecosystem, but also helps payments companies protect themselves from losses due to fraudulent merchants in a way that was impossible by manual review.
In the hands of payments industry underwriters, AI and machine learning can be a very helpful tool for fighting fraud. It’s a focus for payments professionals.
Best of all, AI solutions for fraud are on the back end, so the payments experience becomes frictionless, convenient, and more secure for the payments industry’s customers — consumers and merchants.
Powering Better – and Cheaper – Customer Service
Perhaps one of the biggest areas that AI can improve is the customer experience for payments companies and financial institutions.
Enter chatbots — AI programs that use natural language processing technologies to conduct a conversation. Applied to customer-facing settings, these programs have the potential to disrupt many service industries by offering customized and personalized service in a highly automated, highly scalable way.
The presence of chatbot programs will save banks billions of dollars in operation costs and hundreds of millions of workforce hours when applied to a customer-facing setting like customer service and dispute resolution, according to a Juniper Research analysis. The study puts the figure at $7.3 billion globally by 2023, up from $209 million this year. That’s a staggering 862 million hours saved.
For payments technology companies and financial institutions, AI quickly could become an invaluable tool for resolving chargebacks, helping their merchant customers and streamlining the onboarding of merchants in a very cost-effective way.
Driving Digital Transformation
Like many industries, the payments technology ecosystem has been affected fundamentally by the ongoing digital transformation of our economy. As financial services and retail increasingly move into digital environments like mobile payments, voice commerce and online banking, AI is poised to be an important tool in driving digital interaction for consumers.
Consider that most smartphones shipped in the United States today are equipped with an AI-powered virtual assistant. The smartphone has become the ubiquitous touchpoint for many elements of a consumer’s life, including personal finance. With the proliferation of mobile banking apps — which according to J.D. Power have the highest levels of satisfaction among smartphone users — hundreds of millions of consumers are taking their payments and banking online to their smartphones.
The application of AI in these mobile banking settings makes perfect sense as a driver of the convenience of digital banking. In fact, the dominant channel for chatbot integration will be mobile banking, accounting for 79 percent of successful integrations in 2023, Juniper Research’s chatbot study found. The implementation of chatbots in settings including mobile banking apps will result in a 3,150 percent increase in banking chatbot interactions between 2019 and 2023.
Take Bank of America’s “Erica” smart assistant as a prime example. The virtual assistant served 1 million users in just three months after its launch in summer 2018, according to American Banker. It has remained popular, thanks to its effective use of AI to let users of the app navigate their transactions easily, manage their personal finances, and engage in customer service inquiries with the use of voice commands. It’s a functionality that puts the bank in the same tech arena as many disruptive fintechs or “challenger banks.” Ultimately, it advances the digital transformation of the ecosystem as a whole.
Consider also the growing number of households in the country equipped with a smart speaker or other AI-assistant powered device. One in three owners of voice-activated smart devices already have made a purchase using voice alone, according to Socratic Technologies.
Voice commerce is expected to grow in double-digits in the near-term, based on industry estimates. AI, then, opens a new channel for commerce — one that payments companies can leverage to create more digital touchpoints for consumers to make purchases conveniently and securely.
Finally, AI can drive digital transformation for merchants of all sizes by taking payments data to the next level. For example, machine learning algorithms can analyze transaction data to find patterns — seasonal dips in revenue, for example — and help business owners plan and compensate, down to the most minute decimal point. Further, they can provide targeted marketing capabilities like rewards programs and analytical dashboards to help business owners manage their inventory, capture new sales, and optimize their businesses for each consumer.
Taken as a whole, AI holds much promise for payments technology companies by being a valuable tool to provide a more powerful payments products, by driving consumers and merchants toward more digital commerce opportunities, and, most importantly, by creating a safer and more secure ecosystem.