AI for AML: building Human AI

What is the benefit for human beings, and thus businesses, that AI technology creates, and how do we maximize it?

Theresa Bercich
4 min

Welcome to the second part of our three-part blog series on Human AI! In the previous post, I explained our main vision for Human AI: separating the responsibilities of modern AML and assigning workloads according to the respective strengths of humans and machines.

In this article, I want to go deeper into the technology of (Human) AI, and dispel some misconceptions and misunderstandings about artificial intelligence.

AI: a technology like any other

Artificial intelligence is no more or less a utility to be leveraged than, say, the internet or cloud computing. But like any other new technology, sensationalism can often exacerbate negative perspectives, painting much larger problems than actually exist.

First thing’s first: rest assured, even if misunderstood and misapplied, an AI will not take over your company. Or the world. AI is not Skynet.

On a more serious note, however, poorly designed or deployed AI will make your investment lose its value quickly. You pay a lot of money for supposedly "AI-powered" tools, it’d be bad to find they add to your problems instead of solving them.

AI can be immensely helpful when used correctly. Innovation is a way for information and knowledge to be restructured in more efficient ways. And so, just as the printing press didn't kill storytelling, video didn't kill the radio star, and the internet didn't kill television, AI will not destroy (replace) humanity. Instead, technology allows our human strengths and knowledge to, like nature itself does continuously, evolve.

The way to achieve this stable and reliable benefit of AI is to rethink its utility and the value we expect from it. And that, I think, is the key: what is the benefit for human beings, and thus businesses that technology (AI in this case) creates?

Balancing the promise of AI with its value

A recent survey on AI in the financial industry reveals high hopes for leveraging artificial intelligence as a key factor in future success. 83% of overall respondents, and 81% in the executive C-suite, say they see AI is important for their company’s success. Looking for “increased revenues, reduced operational costs, and greater customer satisfaction,” financial institutions look to AI for a competitive advantage.

Such ambition for AI not only opens opportunities for solutions vendors but also places responsibility on them to deliver a return on investment – or why should businesses deal with the procurement and deployment of AI-driven solutions?

Using AI in the AML compliance space speaks to all three of the above-cited benefits:

  • reduced operational costs by improving productivity and increasing efficiency in compliance processes
  • increased revenues by moving compliance from cost center to growth and expansion enablers
  • greater customer satisfaction by making compliance checks and customer-facing processes frictionless

​​These three core benefits vary in scale and timeframe, and depend on the scope of the implemented AI, but they all share a reliance on the design of the AI. That’s where Human AI solves problems other AI solutions often face.

The race to find dirty money

As digital transformation progressed and financial crime went global and sophisticated, the amount of variables to consider became both too big and too diverse for humans to process. Compliance workflows became infinitely complex, and tools no longer suit analysts trying to evaluate and process cases using disparate and disconnected data silos.

A way to illustrate this is thinking of compliance workloads as decision-making in automotive racing.

Racing teams from Formula 1 to endurance rallies are all deploying data-driven decision-making because of the staggering amount of variables they have to balance. Between factors like the drivers, the track, and the car and its own byzantine balance of variables from tire type to aerodynamics, teams need to use advanced data processing to predict outcomes with a great degree of certainty and accuracy.

Racing teams need to constantly adjust to the race and its conditions meaning they need to understand why things change and what the best decision available is. Everything has to work together, and further: everything plays out over time, with a single race no longer decisively determining the final outcome.

AML investigators are in the same position. They need to consider not only the actor itself in a transaction, but the location or industry they’re in, along with data points such as network, financial products, and shifting behavior — all needing to be placed in the larger context of societal changes (such as the COVID pandemic) and considered over time.

AML solutions need to provide support for processing this multidimensional data matrix fast and properly. Technology like AI is capable of delivering a highly accurate evaluation of data. But it needs to be humans that make the (informed) decisions, and data needs to be presented in ways that are made for humans.

That’s why Human AI is the best way forward for AML.

ML against ML: machine learning to root out money laundering

Human AI uses the multidimensionally scalable processing power of AI but also takes into account how it is presented to the human professional, using intuitive visualizations that represent the new way of thinking about behavior and contextual data, instead of standalone pieces of information or a packaging them into opaque machine-made decisions without transparent reasoning.

Human AI processes the data and makes predictions like an AI would do — but it points to an easily understandable direction, instead of creating an obscure verdict. It ultimately lets human insight and understanding take over the decision. This way, both the machine and human elements are empowered to do what they do best.

Explain, build… deploy!

In the third and final post in this series, we’ll look at the impact Human AI makes on productivity, efficiency, and extending the value contribution of compliance teams.

A crucial part of using AI right is deployment and onboarding. Even the most robust AI becomes worthless if it stays unused. Many approaches skip the human element, the direct beneficiaries of AI technology – but it’s core to Human AI.

In the meantime, let me know what you think about the opportunities of AI, how you think Human AI could assist your business to do AML compliance better, or if you have any other questions!

Sign up for insights from Lucinity

Recent Posts