For most banks, the funding of serious crimes such as human trafficking, drug smuggling and corruption is a continuous fight. With ever-changing regulation to be aware of, technology is leading the fightback for financial companies across the globe.
Working to combat this is Theresa Bercich, Principle Data Scientist at Lucinity, an AML company that utilises human and artificial intelligence to help banks deal with the ever-evolving threat of money laundering. Moving to Iceland in 2018, Theresa was among the first two employees at the company after studying business management, learning to code in several languages and completing a master’s degree at University College London with a focus on machine learning.
Beginning her career as a contractor to start-ups looking to integrate learning technologies into their business, Theresa now uses her skills to support financial institutions to better protect themselves from money laundering and fraud. Here she discusses money laundering, and why AI and humans need to work together if banking is to advance beyond it.
You may not be surprised if a developer of advanced artificial intelligence (AI) software tells you that human beings are fallible. ‘To err is human’, after all. It means systems created by humans are prone to human problems. And those problems cannot be solved in isolation by human minds.
This is true of anti-money laundering (AML) systems, put in place by considered, motivated and well-meaning compliance teams at banks and institutions, who have devised them to protect our financial system. The heft of such an intensely-regulated environment, however, has meant it is slow to evolve. And so these rules and regulations, usually monitored solely by humans or outdated tech, can be recognised and tricked by them as well.
When a regulator can’t miss anything – when banks have to manage resources, meet shareholder requirements and remain profitable – they must leverage new technologies to lighten the burden of their responsibilities. The good news is that for decades, we have witnessed an exponential increase in technological capabilities and the refining of artificial intelligence. The bad news? Crime evolves, too.
In fact, the intricacy of crime and the complexity of regulations has now reached an inflection point, one magnified by Covid-19. The pandemic sped up our acceptance and adoption of new technology – and an acceleration in the digitisation of money, in which illegitimate actors found new places to hide. This ‘new normal’ exposed outdated legacy systems that were not able to keep pace with such a rapid shift in human behaviour. Covid-19 showcased how better technology can make the life of AML teams a lot easier.
It all starts with relationships
Banks know a thing or two about relationships. They have millions of them – all of which rely on trust, vetting and verification. They know better than anyone that humans are not always what we say we are, but what we do and with whom we connect. It’s the analysis of these relationships that uncovers anomalies indicating fraud, money laundering and financial crime.
Human beings are only capable of managing up to 100 (at best) relationships in any in-depth capacity – and therein lies the problems for compliance teams and investigators. AI software, on the other hand, doesn’t look at AML as a monolith, it looks at specific behaviours, starting with one customer to pan out – and from this analysis of thousands of decisions, transactions and connections a new, unified and holistic view emerges.
But just like humans, AI also has limitations. Data in a vacuum needs a human touch to contextualise it. The solution is a balance between the two, to combine the power of AI and intelligence and innovation of human problem-solving. Human AI is what the sector needs to take advantage of.
Empathy in information
It’s no secret: detecting money laundering is time-consuming, expensive and difficult. In fact, over 95% of system-generated alerts are closed as “false positives”, and only 2% of cases actually culminate into a suspicious activity report. However, by understanding the pain points, AML solutions via AI can dramatically improve the current hit rate.
Human AI starts by looking at an actor, building a profile of past behaviour and changes over time, and vectoring in on transactional networks which finds complex hidden representations that humans alone can’t uncover.
Once the system flags suspicious activities, an AI advisor provides investigative teams with risk scoring, case summaries and data visualisations. This is possible because our software, rooted in industry expertise, is intuitive and explainable. After all, it’s one thing to apply AI, it’s another to translate what the AI finds for compliance teams.
During the process of co-development with key banking partners, we found multiple significant tangible benefits such as increased capture rates and true positives, reduced false positives, slashed operational costs and faster case completion. A fintech partner saw a fivefold increase in capture rates and case speed, and efficiency was on average 12-times greater. And because inefficiency makes everything more costly, improving compliance systems means everyone benefits.
Progress in partnership
Outside of the industry, and even sometimes within it, AI can be regarded with suspicion as a dark force that will replace human potential. We must take a different view, one in which AI is a tool that enhances human intelligence without supplanting it.
And never has that more relevant than this year. With the backdrop of geopolitical unrest and the increasing threat of Covid-19, many experts believe banks can also expect to see investigations and financial penalties in relation to misconduct during the coronavirus pandemic. This is why the financial industry must combine its best people with the right emerging technologies to ensure real progress is achieved and heavy fines are avoided.
When we look at the decade ahead, there will be no shortage of problems, be they criminal, political or environmental. Yet by ushering in the age of Human AI, by combining human and machine ingenuity and insight, there will be no shortage of solutions.