Amid all the other emerging and morphing technologies that banks continue to have to deal with, it looks like they’ll have to add another to the list: artificial intelligence.
Based on observations from a number of analysts, the rise of artificial intelligence applications in financial services has gone far beyond facetious comparisons to movie-borne rogue computers and androids that seek to destroy mankind.
AI applications now seem more likely than ever to be applied to everything from battling money laundering, to making loan decisions, to individualized financial planning, to regulatory compliance, and more.
More wifi than sci-fi
Really. And it makes sense. After all, while robots have been building cars for decades, soon it’s likely that they will be driving them as well.
How much further of a leap is it, then, to use what Aite Group calls “inorganic intelligence” to take computing to the next level, further increase operational efficiency, and improve results?
“Inorganic intelligence has the potential to be instrumental for streamlining client onboarding, mitigating and identifying cyberthreats, and developing more intuitive marketing analysis and operations analytics,” says David Weiss, Aite Group senior analyst.
Aite takes this even further, defining what it means by “inorganic intelligence.” Briefly, that refers to the combination of many different but related disciplines: cognitive computing, automation, natural language processing, cloud computing, digital labor, big data, distributed networking, complex event processing, machine learning, data mining, robotic process automation, and concurrent/parallel/distributed computing.
Whew, what a list!
For now, we’ll just stick with AI, although it is interesting to get a sense of how all these technologies are related.
Auto driving change
The reference above to automobile manufacturing is particularly apt in light of a blog posted in Finextra by Graham Seel, principal with BankTech Consulting. He condensed a conversation he had with Nikunj Mehta, CEO, Falkonry. They had talked about the similarities between financial and industrial operations, in relation to AI.
They note that both industries must manage extremely complicated environments of risk, and that AI technologies apply to both industries in these areas:
• Data-informed operations as the basis for day-to-day operations.
• Alarm fatigue—that is, guarding against the expectation of false positives at the risk of allowing real threats to sneak through.
• Trend spotting—Dealing with complex, fast- (or slow-) developing, and nonintuitive trends early enough to take action.
• Smarts that grows—Systems that not only are expert, but are capable of learning.
• Compliance maven—Not only comply with regulations, but are able to explain to examiners in understandable ways how compliance is achieved.
• Protection of trade secrets.
“We see an opportunity to use industrial process exception management techniques to address some of the most challenging issues in bank operations today, including (among other potential areas) fraud detection and anti-money laundering,” Seel concludes.
AI can know your customers
Which is interesting because Celent just came out with a report saying the same thing, and mentioning AI by name.
“Traditional rule-based KYC-AML technology necessitates significant dependence on manual efforts particularly in alert investigation stage, which is costly, error prone, and inefficient,” says Arin Ray, an analyst with Celent’s Securities and Investments practice. “AI-enabled solutions can not only automate significant parts of operations but also offer superior insights through advanced capabilities for analyzing structured and unstructured data.”
Dan Schutzer, senior technology consultant for BITS, writes that AI has been around since 1950 when Alan Turing proposed a test of a machine’s ability to exhibit behavior indistinguishable for that of a human (the famous “Turing Test”).
Since then, interest in AI has been up and then down at least twice, through the mid-1990s. Now, though, things have changed, he says, including the declining cost of computing and the increasing power of computing.
He notes that technology companies already have developed algorithms that track a user’s online habits, creating deeply personal online experiences.
“The implications for the financial sector is that by tracking users’ habits, activities, and behavioral characteristics, financial data and products can be personalized to meet and anticipate each user’s unique and changing needs,” Schutzer says.
Specifically, he points to these likely areas of applications in personalized financial services: automated financial advisors; digital and wealth management advisory services; smart wallets; insurance underwriting; loan decision-making; trusted financial social networks; and more.
Machine learning and self-service
IBM’s Paul Davis, writing in that company’s banking blog, waxes even further about how AI could boost focus on individual customers.
“Consider new ways to interact with clients via digital virtual agents where machine learning meets self-service,” Davis says. “In cognitive banking these new cognitive systems can interact with customers, listen to questions, and offer solutions. They learn with every human interaction and grow the collection of knowledge, quickly adapting to the way humans think.”
[Note: “Cognitive banking” goes back to Aite’s mention of “cognitive computing.”]
Further, Davis expands on the regulatory compliance aspect of AI: “By understanding the latest regulations, and comparing bank activity against them, cognitive systems can help the bank’s risk officers detect when a bank cannot meet its obligations. They can also provide recommendations on how to correct course.”
Still further, Davis remarks on security: “On the security front these systems can understand, reason, and learn about security topics and threats tapping into security knowledge that has previously been dark to an organization’s defenses. These systems enable security analysts to gain new insights and respond to threats with greater confidence at speed and scale, performing statistical analysis on a corpus of data, looking for patterns, defining relationships between that data, and forming hypotheses to help people make better decisions.”
Looking at “the other hand”
On the flip side, though, BITS’s Schutzer does point out serious business and privacy issues that will accompany the onset of AI. Just some of these issues include:
• How does a user distinguish one automated online banking application from another?
• How can one benchmark and rank the quality of recommendations?
• Will more comprehensive access to data across institutions result in better advice?
• How easy will the system be to use?
• What happens when security and fraud protections fail?
• How can user privacy be guaranteed?
Nevertheless, he concludes this way: “Because of the significant potential benefits there is probably no turning back, there will be increasing automation of financial services, often employing AI technology. However, these new AI applications introduce a number of business, security, and privacy issues which will have to be addressed if they are to succeed in the marketplace.”
In the meantime, I will just tell my self-driving car to stop at the ATM in my bank’s drive-through lane. I hope it wakes me up when it gets the cash and doesn’t splurge it all on high-test.
Sources for this article include:
- Adopting a Culture that Will Prevent Algorithm Bias
- Addressing an Apparent Contradiction in Credit Scores
- Stanford Federal Credit Union and CITI Partner with Google
- Banking Algorithms, the Apple Card and Sexism
- Senior Official Recommends the Launch of a Real-Time Payment System to the Federal Reserve