Artificial Intelligence and its Role in Compliance – How the Machine Can do Better

Raffael Maio, Managing Director APAC and Co-founder of Swiss fintech NetGuardians, believes that the machine can do better. What he means by that is that AI can augment and improve on the current protocols generally employed by banks to root out non-compliant transactions and clients. He expounded on his theories at the Hubbis Compliance in Asian Wealth Management Forum in Singapore.

Maio began by reporting that banks spend some USD270 billion globally each year on compliance. “Banks are paying huge sums just to keep up with regulation, just to be able to almost do what the regulator is asking,” he explained. “That is not going to change in the near future because the pace of technology is such that more and more problems emerge. Indeed, some institutions are already spending more than 10% of their budget on compliance, or even more. For a major US global bank, that, for example, represents more than 20,000 employees.”

Running fast to catch up

Maio’s premise is, therefore, that to continue along the same route is no longer advisable because the financial industry cannot keep pace. “The main limitation of this industry today is the lack of scalability, it cannot keep hiring people endlessly, infinitely in order to solve the challenges of all the new rules that keep coming, non-stop.”

Maio observed that while the hiring of people in this area is affirmative, it is also immensely time-consuming, costly, inefficient and ultimately not sustainable. “Particularly as the bulk of the effort is to tackle false positives,” he added, “in other words false problems. Moreover, the more people, the greater the risk of mistakes, and not actually in real time. This is why banks are embracing technology; they must move faster and more efficiently, especially in a world where instant transactions are the new norm.”

The machine can do better

And here, Maio stated, is where AI can make a dramatic difference. “Financial institutions have already made great steps forward to become ‘smart’, for example in the area of customer experience, robo-advisory, personal advisory, which is where AI is playing a growing role. In China, for example, almost everyone seems to have a device with a robo-advisor in their pocket. And we can extend this AI role into the area of compliance? Well, it is taking place now, we are witnessing the transition from an era of manual processes to the age of digitalisation."

Maio expanded on this by explaining that rules-based protocols are no longer valid. “When there was far less data, the rule-based system was acceptable,” he noted. “But the complexity has magnified exponentially, and the rules-based system creates far too many false positives and as a by-product has created financial impact for some institutions and reputational damage, as well.”

Throw away the rule books

But AI offers the opportunity to use the enhanced computer power available today to handle the vast proliferation of data. “We can analyse several years of information in moments, whereas ten years ago it was simply not possible, logistically or financially. But with new AI solutions today, which also learn by themselves, by the way, we can in real time achieve the results we all seek, away from the static rules-based systems that are now defunct.”

Maio explained that the machine can learn about the habits of individuals, employees and other and detect suspicious transactions. It can analyse transactions over several years; it can build dynamic profiles and then keep those profiles up-to-date in real-time. It can then compare transactions with the customer/user profiles and thereby compute a risk score and take a decision.

Dynamic, real-time, evolving

Real-time solutions that can automatically change dynamically are therefore the way forward, Maio believes. “As we can now compare transactions in real time from groups, from customers together, from individuals and thereby compute more accurate risk scores, we are making the system and process far more effective and cost-efficient,” he maintained. “Now, with this advanced AI, we can see where there are deviations amongst individuals, corporates and so forth, in relation to their peers. We have a far bigger picture, a broader vision,  and we can truly compare on multi-dimensional axis, we can learn.”

He gave an example of someone like himself buying a car. “I do not want the bank to block my purchase because it is a single large transaction, so for the bank to tackle that false positive is a waste of time and effort. The banks need to be able to assess against the individual and their peers so that they become far more operationally efficient.”

Maio concluded by contending that AI is neither a dream nor a threat. “AI should today be seen as a tool to help individuals, to help them achieve their goals faster, cheaper, and more efficiently. By the effective use of augmented intelligence, it will, I believe, achieve better results without the danger of putting humans out of work. We can co-exist and do so very well.”