Artificial intelligence gets real: The new era of AI

Four characteristics that make today’s AI more intelligent and less artificial.

It’s fascinating to follow the advancement of AI. What started as sci-fi material and became a vague buzz term is now a practical tool that improves our lives in countless ways. By 2030, 80% of today’s emerging technologies will include AI capabilities, and even today, 77% of consumers use AI-based technologies (often without realizing it). AI brings many impressive advantages to a long line of fields, very much including money laundering prevention. Here are four characteristics that make today’s AI more intelligent and less artificial.

For the people: Human AI

Research by the Harvard Business Review found that companies reach the best performance improvements when humans collaborate with machines. These man-machine partnerships improve the abilities of both parties and create more considerable and unique value.

Human AI has different meanings. First, it refers to AI that is people-focused and studies complex human behaviors. Then, it includes the collaboration we’ve mentioned here, where humans’ active contribution is needed to complete the picture, and algorithms alone are not responsible for the entire process. Finally, human AI is created for people’s benefit, not against it, and designed to serve customers and employees.

In the anti-money laundering (AML) field, human AI also signifies the shift from a rule-based approach to a risk-based behavioral one. Lucinity’s detection mechanism studies customers, not transactions, and builds a comprehensive profile to detect suspicious patterns. Our approach relies on continuous feedback from human investigators that learn from the algorithm and teach it simultaneously. This symbiotic relationship makes all the difference.

Find meaning: Explainable AI

A recent IBM survey found that 83% of companies consider the ability to understand the algorithm and learn how it reached a specific conclusion critical. Companies rightfully want AI to teach them how to fish.

Once again, the notion of explainable AI carries multiple meanings. It refers to a solution that is easy to understand and offers clear insights that turn into actionable decisions. Instead of spending a great deal of time figuring out what the algorithm’s data means, companies want AI-based solutions to offer this information.

Then, businesses want to understand the model at the basis of these decisions. Democratizing this knowledge gives companies more power and allows them to make the most out of this incredible technology. AML algorithms will not only flag problematic behavior but also explain what makes it stand out.

We have had countless conversations with AML professionals around this simple observation: “I want my team to spend more time making decisions and less time making sense of data”. Explainable AI addresses exactly this issue.

In the AML world, explainable AI plays a third vital role. It enables financial institutions to provide clear and detailed explanations to regulators for ongoing procedures and audit purposes.

Forget one-size-fits-all: Contextual AI

Context is what puts “intelligence” in “artificial intelligence”. We need to teach algorithms how we interpret different situations, especially when complex behaviors are in question.

AI must study behaviors in light of the surrounding situation. Certain actions may seem innocent enough outside of context, but when connected with other behaviors and insights, they become highly suspicious. This is part of both explainable and human AI, because it forces us to be focused on the actor and offer accurate explanations as to why the algorithm chose to flag this transaction.

Contextual AI should also be modular. Lucinity’s solution allows businesses to tailor the AML platform based on specific conditions and needs. Companies offer their own unique context and the system studies it and operates accordingly.

Better together: Open-source AI

When it comes to implementing AI technology, one of the most significant obstacles standing in businesses’ way is data collection. Data provides the building blocks that feed AI algorithms and continues to improve it over time. Without a sufficient amount of data, there is no foundation for AI operations. Gartner finds that gathering data is one of the top three challenges of AI adoption. McKinsey agrees, stating that only 18% of companies have a solid AI data-sourcing strategy.

The solution is to form an open-source AI community, where companies share their insights in order to grow together. In the fight against money laundering, where we focus on interpreting sophisticated scenarios that could otherwise lead to AI failures, open-source solutions are a must. They put things in context, offer multiple perspectives, and allow us to innovate faster and better. Money laundering is a global social problem and the international financial community must join forces in battling it. Like Lucinity, tech giants such as Amazon, Google, Facebook, and Microsoft opened many of their AI systems. We call on other companies in the field to join this effort.

We practice what we preach and design our technology around AI that is human, explainable, contextual, and open. What may seem like added value is really the fundamental pillars of AI, which will make it more effective and usable. If your AML technology lacks one or more of these characteristics (or doesn’t offer AI capabilities at all), contact us and learn how you can take things to the next, more intelligent level.

Sign up for insights from Lucinity

Recent Posts