–> This is a pre-event interview in the run-up to the Leaders in Finance AML NL Event on 5 October 2023.
Jeroen: Ali, thanks for taking the time to speak to Leaders in Finance in the run-up to the Leaders in Finance anti-money laundering Netherlands event on the 5th of October this year. We are very glad that you will take part in the event. And we are also glad that you are taking the time now to speak with us. First of all, maybe you could shortly introduce yourself?
Ali: Thank you. I am head of Data at bunq. My goal is to enable the company to implement data-driven solutions and make sure we, as a tech company, stay on top of the latest trends in technology. One big focus of mine is making all of our processes better with the help of AI & ML. As such, I’ve been working on our transaction monitoring system that’s proven to be 2.5 times more effective in detecting fraud than other means used by the majority of banks and financial institutions. I’m now working to implement AI in a multitude of functions at bunq, such as user support, marketing, or app testing.
Jeroen: Great. I think your background is something that is highly in demand, isn’t it?
Ali: Yes, I would say so. We get a lot of requests for talks about AI, and there’s a lot of interest in AI in general, especially after the emergence of ChatGPT and after bunq’s court case with the Dutch regulator about the use of AI for transaction monitoring.
Jeroen: Given you have been working on this transaction monitoring system for bunq, I guess you have also seen a lot of change over the years. What do you think are some of the most important changes you’ve seen related to AML?
Ali: From the perspective of a data scientist, we see that types of approaches used to do transaction monitoring have evolved over time. We moved from very traditional, very simple rule-based systems to more complex models, both AI-driven and statistical model-driven. That is now being embraced by larger financial institutions. Ever since the Dutch regulator started embracing AI, a lot of banks have started adopting this technology, so that the banking system could become safer for everyone. So that is one of the trends I have seen. Also, the amount of data that we have at our disposal nowadays to learn about thought patterns is increasing rapidly. I am now dealing with hundreds of millions of transactions. So the predictive power of the system we are building is higher every day
Jeroen: Are there particular things that you see unfolding right now? You already kind of mentioned it, but are there things that you see going to happen or developments that are going to unfold in the next few years?
Ali: Yes, definitely. One of the things with AI was always: it is a black box, how do we explain what it is doing? Nowadays, we can get a transaction monitoring hit, and a model can predict that it is potentially fraudulent. Potentially, because a human has to investigate it as well. And with the help of a language model, you can now explain why exactly. I am seeing a lot of companies that offer transaction monitoring solutions already adopting language models. They make the whole process much easier, much simpler to build because it is language. Humans are driven by language. It makes things more interactive. You can actually talk to a model instead of just getting a prediction, for example when you’re asking ChatGPT to generate text or solve a programming problem. For me, that is the main area we are moving towards if I would have to make a prediction based on what I am seeing today.
Jeroen: That is great. Makes total sense to me. Are there things happening currently in the AML landscape from a data or banks perspective that you would like to see happen? You referred to the court case, we have done a podcast recording about this, it was one of the best listened to podcasts in a long time, with the lawyers who were your lawyers versus the DNB. Are there particular things that you would like to see happen from your data perspective or bank’s perspective?
Ali: When talking about our transaction monitoring system, I tend to always explain there are ‘known knowns’, ‘unknown knowns’, ‘known unknowns’ and the ‘unknown unknowns’. And the biggest problem for data scientists tackling transaction monitoring is ‘unknown unknowns’. Something that we do not know might be a ‘known known’ for another bank. So collaboration can be better for everybody. There are initiatives like Transaction Monitoring Netherlands that helps unite forces within the financial system and tackle financial crime together. Collaboration is key in this area because when you collaborate, you can actually be more effective as a system.
Jeroen: You would probably agree with this statement that it takes a network to fight a network?
Ali: Yes, definitely. Network analysis is something that we use as a technique, but it always ends where our data ends. And then it is the other bank doing their stuff in a siloed manner. And if you just add one connection, it becomes much more effective.
Jeroen: That will depend on a lot of laws and regulations like TMNL, as you mentioned, if that will actually come to fruition. Do you have tips for people related to AML?
Ali: Yes. One thing I learned at bunq was: do not get stuck in overthinking, and just test it. Go to the point where you come up with a solution, test it as quickly as possible and then get feedback from the environment, from the system itself and use that to improve. Instead of spending a lot of time on theory only to figure out that the approach might be flawed. This approach is something that clicked with me when I started working at bunq.
Jeroen: That is great. Is that a tip for a specific group or for everyone?
Ali: I would say it holds for pretty much everyone. You can apply it to whatever you are doing. We tend to spend a lot of time overthinking or thinking about theory, while in practice things may be completely different.
Jeroen: At the event on the 5th of October, we have banks, we have a newer bank like yourself, we have academics, we have the prosecutor, we have the Dutch Central Bank, everyone is there and the whole ecosystem of the fight against money laundering. If you go to an event like that, what are the things that you would like to get out of it?
Ali: First of all, I would like to see the perspectives from different people on the topic of anti-money laundering in general. So how do people view this from their own unique view? So a data scientist or a regulator or someone with a legal background, for me, that is super interesting because everybody has a unique view on things and in the bigger picture it is useful in some way or another. Secondly, I just try to see: Where are different parties at in terms of technology when they are using AI or data science? What are they doing? What are they using? Certain parties are usually much more ahead in terms of network analysis, while others have streamlined their processes for throwing out new rules very quickly and very well. Learning about other companies’ experience can be a great source of inspiration.
Jeroen: Great. Ali, thanks so much for taking the time to talk to us. We are very much looking forward to having you on our AI panel. You are definitely in the right spot to add value there. Thank you so much.
Ali: Thank you.
–> This is a pre-event interview in the run-up to the Leaders in Finance AML NL Event on 5 October 2023.