A&M's Peter Kwan shares his perspective on the release of (another) new AI-based AML Compliance platform:
The 2015 acquisition of a natural intelligence platform by a global computer parts manufacturer has just identified its first problem to tackle. In an unlikely alliance, the newly branded Artificial Intelligence (“AI”) platform has reemerged as a contender in the already crowded Anti-Money Laundering (“AML”) RegTech space based on a company issued Press Release last week.
As the article mentions, the new AI platform is based largely upon the inner-workings of the human mind. It applies a method of "continuous learning" to all the structured (i.e. data) and unstructured (i.e. documents) information you can throw at it. Interestingly, this is just the latest in a series of major investments made by similar technology companies looking for a way to prove they can be the best-in-breed AI solution for the AML marketplace. Other forms of AI have also surfaced in the last 5 years offering promises of a smarter, better and leaner AML department by way of Cognitive Computing, Neural Networks and Supervised Learning.
On one hand, AI promises a more efficient workface, a reduction in bloated compliance budgets and the promise of next generation cognition that will allow organizations to tune out the noise of false positive alerts. On the contrary, AI also poses a set of severe challenges for organizations that take the leap of faith. Moving from simplistic rule-based algorithms to complex algorithms that have the potential to make decisions without any human intervention (and even learn from its own mistakes) raises the bar for what can be understood, replicated and explained by even the most astute compliance professionals. This learning curve from within, coupled with the professional skepticism of regulatory bodies from without, poses an immense challenge for AI adopters (and attempters).
While as a technologist I am hopeful of what possibilities these technologies might bring, the professional skeptic in me thinks there's still a lot of road ahead. Only time, testing and the elasticity of AML compliance budgets will tell how the new breed of bigger AI platforms will fare.
It is said to mimic the associative memory of the human brain to surface similarities and anomalies hidden in heterogeneous data sources, while accessing larger data set than its human counterparts.