The Case for “Genuine Intelligence”
Or "Why AI isn’t enough when it comes to preventing financial crime"
People like to say that “We don’t like change”. This isn’t accurate. When a new, game changing technology becomes available, it is usually embraced by the hordes of “Early adopters” and closely followed by the “Early Majority”. I think a more accurate statement would be, we do not like gradual change. Anyone who’s ever woken up to find their smartphone has updated itself can attest to this.
This is even more evident when talking about compliance, where most organizations either put off making any changes until some imaginary quarter when they will “have the bandwidth” or want to implement the most advanced machine learning algorithm in the world yesterday.
Unfortunately, both approaches are flawed. Putting off improvements to the efficiency of your operations has a huge cost with no benefit since eventually you’ll have to do it anyway. While jumping head first into a new and untested technology requires large resources and carries with it a high chance of failure.
My view is that the right compliance platform needs to combine elements of the old with elements of the new. On the one hand, a rule-based system for flagging suspicious activity is necessary because this is still the language the regulator speaks, and they are the ones who write the rules. On the other hand, manual processes are inefficient and inaccurate and, as has been proven in many industries, should be replaced by automation.
To properly utilize Machine Learning based detection, the algorithm must be “educated” on what to look for. This is why Clarus uses various techniques to identify suspicious activity, with a rule-based solution serving as the base and a high level of “data labeling” providing the building blocks for using machine learning effectively.
Also, compliance is not a uniform problem and cannot receive a “one size fits all” solution. Any technological solution needs to understand the many money laundering and terrorist financing schemes and the specific risks an organization is subject to. For this reason, at Clarus, the on-boarding process of a new customer, which customizes the platform to their needs, is critical to them successfully getting the most out of our platform.
To sum it all up, while AI and Machine learning are definitely the future of the industry, they are only an engine, and an engine without a steering wheel, brakes, and a well-trained driver, can cause more harm than good.