“What we need is a psychologist rather than a technologist.” That was the comment made by a senior compliance leader from one of the world’s leading banks at the recent 1LoD Summit in New York City.
With it, she highlighted just how different the future of surveillance might look compared with current practices. I may only be weeks into my new role at Digital Reasoning, but the dynamism that artificial intelligence (AI) is bringing to conduct risk mitigation in banking has quickly become apparent.
New ways of thinking
The industry-changing impact of next generation communications analytics was brought home to me at the 1LoD Summit, where over 200 senior surveillance leaders represented around 70 firms. All 3 lines of defense were present and I can’t recall a more focused and productive industry gathering. It provided a unique perspective on how banking’s risk communities are adapting to changing expectations and new technologies.
Among these are new, AI-enabled capabilities to identify complex behavioral indicators, such as language signaling intent. Legacy lexicon based technology resulted in the costly problem of excessive false positives. AI-enabled Natural Language Understanding and analytics helps to resolve this, but as the compliance leader I referred understood, the real benefits will only be found with new ways of thinking about conduct risk.
Systems like our own Conduct Surveillance, now used by the majority of top ten banks, offer surveillance analysts the means to resolve analytics insights to individual employees. Rather than working through a list of discrete alerts, it’s now possible to see how accumulating these can build a risk profile and score. The approach is not dissimilar to credit risk scores, which can be tracked over time and changes in the score can be linked and weighted to specific actions.
The difference is that communication is not quite the same thing as behavior. It’s a powerful indicator, but until recently all that technology could do was surface signals in language. Now it is becoming possible to assign a numerical value – an inference of risk from insights captured from multiple communications channels and over time.
Surveillance leaders can see the benefits of this approach. It’s employee-centric, offers a more balanced view, adapts to ever larger communications data volumes, and cuts down analytics silos. It brings a new layer of automation to the surveillance task that allows analysts to spend more time on investigating possible breaches of regulations and codes of conduct, rather than wading through alerts.
Regulators say they want to see this progress; banks are becoming capable of delivering it. The challenge is establishing trust in new ways of understanding data and its impact on working practices if industry wide adoption is to take place.
This is a fascinating contemporary debate that I believe will have outcomes that impact beyond banking. Digital Reasoning is working with its partners and banking communities, including those run by 1LoD, to increase understanding of the state-of-the-art with AI and relate that to conduct risk needs and challenges for the industry as a whole. I would welcome hearing from anyone in the industry – institutions or regulators – with an interest in this debate. You can reach me via LinkedIn .