AI is a compliance officers’ best friend

AI is a compliance officers’ best friend

By Yannick Martel, Founder, Bleckwen

Interview ahead of Payments Forum 2019  (Get 15% discount on the Forum using presenter code: PAYMENTS88)
Can you please tell the Risk Insights readers a little bit about yourself, your experiences and what your current professional focus is?
What, for you, are the benefits of attending a conference like Payments Forum?

I am Yannick Martel, I am a happy father of 4 kids. A fun fact I can tell about myself is that I like starting new things: I started learning the piano 6 months ago, as I started Bleckwen 2 years ago.

From a professional perspective, I have worked in IT and Data science for over 20 years. From my experience at major companies, software vendors and consulting firms, I have built a broad outlook on solving business problems with innovative technologies. I am convinced that the combined use of Data Science and Machine Learning along with sound software engineering is the new disruptive force for businesses and communities. Since 2016, I have been developing the Bleckwen initiative, applying Data science and Machine Learning to combatting financial crime, in order to make the world a safer place. I am building a team with a variety of expertise to make this vision a reality.

How can sanctions filtering be made less of a daunting task for compliance officers?

Sanctions filtering is always in the mind of the Compliance officer because of the threat of 9 BN$ fines. In sanctions filtering, failure is not an option. We have come to the situation where it is less risky to have armies of hundreds of people dealing with dummy alerts rather than taking the chance of missing something and being fined by the regulator. It’s insane! AI can definitely help the compliance departments in that Herculean task. AI can empower analysts, reducing delays, automating routine activities and even enhancing reliability.

Can you recommend best methods to ensure the value and benefits of newer algorithms are confirmed at each stage through their interpretability?

Newer algorithms needs to be tested on real data, in parallel with older procedures. Then we can review their performance and understand which factors they are using to make their decisions.

When we are convinced of their soundness, we can move them gradually into production. Then interpretability and forensic processes will be very important inputs in the feedback loop to improve them and keep them updated.

Can you highlight some of key factors to why it is important to optimise the mobilisation of operational resources?

It is crucial to optimise the way operational resources manage the alerts generated by the sanction filtering tools. It will have a direct impact on 3 drivers.
1) Cost: improving the process, has a direct Impact on the workload, though on the number of resources needed to do the same job.
2) Operational risk: For a human being, spending 20 seconds per alert, will Inevitably lead to an error. You cannot be fully focused on a task you are going to repeat over 1000 times in a single day. Improving the process will have a direct impact on the quality when allowing more time/alert
3) HR risk: Repeating thousands of time the same action on a low value added task has an impact on human resources and motivation. Improving the process will give more time for investigation though more added value

How can the burden of false positives best be managed?

A straight replacement of existing processes being too risky, tools must be provided for a gradual optimisation of the processing of false positives – keeping manual review of all alerts but optimizing and accelerating the review when possible and having algorithms cooperate with human brains and eyes. This will reduce the pressure and burden on compliance teams and prepare some automated reduction of false positives, gradually.

How do you see the impact of Payments evolving over the next 6-12 months?

Open banking and instant payment will augment the pressure on compliance teams by increasing the volume of transactions, and thus of alerts, and requiring instant or near real-time decision beyond human capabilities – at any cost!

This will make adoption of artificial intelligence a must, else Compliance will be between a rock and a hard place, Marketing on one hand, the regulator on the other.

You may also be interested in…

Make a free account

Don’t miss out on…

X-Tech 2019