The good ‘ole days are usually much better in our memories than they ever were in reality. Even so, it doesn’t veer too much toward the illusionary and nostalgic to submit the proposition that banking, and other financial services, used to be much more personalized — much more one on one — than is the case now.
Sure, mobile banking and payment apps offer the patina and, perhaps less often, the genuine experience of personalization. All the data submitted and left behind online by users serves to give software ever-more information about how specific consumers browse, shop, buy, pay and transact (all of which go a long way to helping prevent fraud, of course).
What if cutting-edge technology — technology not yet in mainstream banking, but arguably headed there with firm speed — could replicate in a significant and meaningful way, the old reality of having a banker who not only has the time to deal with customers one on one, but knows them deeply as individuals? It seems counterintuitive, the idea that algorithms could replicate the customer service and personal engagement offered by old-time bankers (and merchants), but it’s possible.
That idea served as a major thread in a recent PYMNTS podcast discussion between Karen Webster and Brighterion CEO Dr. Akli Adjaoute about artificial intelligence (AI) and payments. A new report entitled “The AI Gap: Perception Versus Reality In Payments And Banking Services” — a PYMNTS and Brighterion collaboration — includes data collected from more than 200 financial executives, commercial banks, community banks and credit unions across the United States to provide a comprehensive overview of how financial institutions (FIs) leverage AI and machine learning (ML) technology to optimize their businesses.
One of the headline findings from the report is that while 100 percent of all FIs surveyed said they use at least one form of machine learning technology, only 5.5 percent of banks have genuine artificial intelligence systems in place. Simply put, AI is capable of unsupervised learning (that is, without human input), while machine learning — no matter how sophisticated, or how much it resembles AI — does indeed require that supervision.
As PYMNTS readers are well aware, the optimism and promise that surround AI (some of it mere hype that will eventually deflate based on actual use cases, of course) are at high levels. Adjaoute said little during the PYMNTS interview to reduce that mostly positive vibe of AI, though he certainly put the technology into a realistic context, free of the marketing buzz feel.
For one, the virtual heaps of data that continue to accumulate in the digital world require more sophisticated analysis, and that can essentially provide job security for artificial intelligence. “The more data we have, the better,” he said. After all, he added, more data — and the capability to spot patterns in it — “allow[s] financial institutions to have a better understanding of who they are dealing with.”
Having the ability to find and analyze data in real time or near real time — data from social media, court documents, SEC filings and other web sources — can also make customer onboarding quicker and more secure, and combines good customer service with good fraud prevention. “Companies can use AI to strengthen the level of insight they have about markets, customers, everything,” Adjaoute said. In short, properly deployed AI can triangulate different sources of data to arrive at decisions about onboarding, authentication and fraud in quick time.
That’s the promise. It’s a big promise — feeling more connected to a financial institution (or other company) via the sophisticated, self-learning computers it has deployed: faceless, bloodless bots that are helpful, not menacing, and which keep people safe while offering extremely relevant product and service interactions.
The reality can be a bit more fuzzy, to say the least.
As the PYMNTS-Brighterion report noted, the term “AI” does indeed function as a buzzword, and has been shaped like taffy to apply to technologies and systems that are actually machine learning. That matters because such misunderstanding can lead financial institutions down a wrong path. Those FIs, after all, have invested billions of dollars in legacy approaches that are largely manual and repetitive, according to the report.
“This includes consultant fees, armies of back-office agents and outdated rules that flag violations of [anti-money laundering (AML)] regulations, which they describe as AI,” Adjaoute said. “These systems have proven to be largely ineffective at actually curtailing money laundering and, as a result, regulators in the United States and the European Union have issued more than $340 billion in fines for non-AML compliance since 2009.”
When it comes to AI (true AI), the early uses are emerging. The report found that 81.8 percent of financial institutions that utilize AI use it for banking services, and 72.7 percent use the tech to fight internal fraud.
Personalization Via AI
AI, as Adjaoute explained during the interview, has the potential to reshape the banking industry and the customer service it offers. That is, it can do much more than identify fake online accounts and account takeover attempts, or see early signs of an emerging fraud ring that might soon cause trouble.
In fact, AI “is all about personalization,” he told Webster. Artificial intelligence can really dig into the behaviors of consumers, and can better determine when a fraudster is trying to impersonate someone, for instance — a judgement that depends on spotting anomalous actions, even at very smaller scales. That provides a stronger defense in a world where, as Adjaoute put it, “criminals are getting smarter and smarter.”
However, fraud prevention is only part of the AI pitch going forward.
The technology’s ability to spot those patterns and anomalies can also translate into better customer service. The data from payment card transactions and web browsing can tell so much to a marketer, retailer or financial institution — for example, when a woman is pregnant because she is buying healthier food and maternity clothes, or when a young adult has just arrived at college. AI, though, can deepen those insights and even make them earlier, providing customer service or retail opportunities for the business using the technology.
In short, AI can spot all those little details and put them into context. In turn, that can lead to “one-on-one service,” as was common with bankers and retailers well before the digital age. “AI can bring back that level of one-on-one service,” Adjaoute said, and help to better determine lifestyle changes among consumers — important information for FIs, as well as other payment and commerce providers.
“Everyone is different,” he said, and everyone undergoes changes throughout life. Staying on top of those changes can result in product and service offers that have a better chance of appealing to that specific consumer at a certain time. “AI takes us back in time, to when the guy who sold you your bread knew you,” Adjaoute added.
As the PYMNTS-Brighterion report explained in depth, “smart agents” is one path through that future, at least for financial services. Smart agent technology is a personalization technology that creates a virtual representation of every entity with which it interacts, including customers and banks. There can be as many smart agents as active entities. That means 200 million smart agents could analyze 200 million cards engaged in transactions — the personalization and security benefits are not difficult to imagine.
However, like human beings already do, AI — to gain more use and stand as true AI — must also be able to adapt to all the changes that happen, even on a daily basis. “Real AI is an AI that can see every single person differently,” Adjaoute said, “but also in a contextual manner.”