Unlocking the Power of AI in Finance: Insights from Transaction Monitoring Models

[Music] good afternoon ladies and gentlemen and welcome to Abu Dhabi Finance week here at the capital of capital ladies and gentlemen this panel highlights a crucial and strategic initiative led by the adgm academy Research Center in partnership with NBD and rabdan Academy together we have worked on an machine learning based alert scoring model that is transforming transaction monitoring processes and significantly enhancing anti-money laundering capabilities today we will dive into the research findings the challenges faced during implementation and the broader implications to the financial sector starting with Zubin the research introduces a Cutting Edge alert scoring model designed to improve the classification of suspicious transactions by leveraging Advanced machine learning techniques how does a model like this improve anti-money laundering capabilities and capacity so if we look at it from a something like transaction monitoring is a bit of a double Ed sword at one end you want to do the right thing catching the bad guys and the suspicious transactions but as we progress you find so many false positives which come out and the amount of time that an individual needs to investigate it Etc is significant the downside of that is you're not being able to report it out promptly when the authorities want to clam down you have a timeline and with the speed of transactions it's imperative that you get it out fast on the other end of the spectrum is how do you do it you don't want to set something very conservative to miss out something uh so you have a bit of a trade-off between Effectiveness and efficiency and that's been a growing concern as transactions keep growing red flags keep growing the output of the systems have grown so what the model is trying to do is prioritization and identification of transactions which based on the model has a higher probability of turning to be really suspicious and therefore you can Target your resources to expedite and handle those thank you zuin it's fascinating to hear how these Innovations are impacting our day-to-day operations in the financial sector uh Dr Eric from an academic lens what makes this project a standout example of translating theoretical AI advancements into practical industry specific Solutions thank you rder I think uh there's three things that stood out for me personally for this uh project the first one was the Innovative approach that was used machine learning and the use of AI in the financial sector and other sectors isn't brand new so to speak it's been around for a few decades now uh but what stood out particularly with this one was the way it was bolted onto existing systems that were already in place so the rule-based system that was implemented by the Banks and the financial sector was able to be added to using this technology so that in itself was an Innovative approach uh but also what also stood out was the fact that it was implement and tested over a significant period of time through a pilot program that was both supervised and unsupervised so what that enabled us to do was to test it in real time uh against uh human analysts to ensure its Effectiveness remained exceptionally high and that's how we got such a great success in terms of reducing the false positives and false negatives uh which was one of the key objectives so it it was it was a standout project from that perspective but then one of the primary things we wanted to produce in in that the white paper has helped us do is produce a blueprint for the rest of the sector and that's what makes this stand out as well because frequently when Academia and Industry come together uh often it's not published in the way this one has been done so we've done a white paper and there's hopefully an academic Journal article on the back end of it as well so uh what we're hoping is that other Industries uh particularly in the financial sector can take this approach and also apply it based on those reports that we've produced thank you Dr Eric your explanation really highlights the value of innovation uh rigorous methodologies and real world testing in creating AI solutions that are not only practical but but transformative uh Zubin I'll turn I'll turn back to you so integrating AI models into existing compliance Frameworks is often a complex process so what specific challenges did Emirates NBD face during this integration and how were they addressed so as Eric mentioned mentioned of course you have ai Etc in the finance world but when it comes to the compliance side for the surveillance there are a bit of a nent stage at the moment it is something where you're testing the waters so at one side when you're migrating there's always a reluctance to give up something which is so-called working at the cost of are you going to miss something how do you get that fine balance of migrating to something and as again it's a Bolton so we had the luxury of of trying and testing that uh what did come up as a challenge is of course training data because there are too many Shades of Gray in this particular vertical it's not a binary and there's a lot of human bias which goes in so how do you really go up and uh ramp it up in terms of the deployment also what comes up is management of expectations uh the moment you he AI ml U expect something on a platter it's a straight through that you're getting one alert and it's going to be a straight s which is not the case and what did need to be socialized and really got in is yes you develop a model you deploy it but then there's a lot of effort in terms of maintaining it so that there's no regression and you don't drop the ball down the line so that governance around the model selling it saying yes you have something but there is a journey which has to continuously be dare to keep a maintenance it's like a car you get it you don't maintain it it's going to really go out so when your insights highlights the operational hurdles that financial institutions face when adopting to AI thank you for this uh Dr Eric based on your experience what do you think are the critical success factors for collaborations between academic institutions and the industry Partners on projects like this one I don't think they change dramatically um specifically for the financial sector I think they're the same with any industry in academic partnership and uh first and foremost they have to be able to align their objectives uh Academia uh their currency tends to be new knowledge research Innovation uh and that's usually manifested in patents academic articles and so on industry tends to have a different objective often uh financially orientated profit margins shareholders stakeholders etc etc so it's important to first find where those two can align uh because that's when the conversations can start uh and then that opens the second aspect which is the need for open communication uh particularly when we were working on this project uh Ai and machine learning they come from a a computer data science background uh and although there's some nuances that overlap with the financial sector they're distinctly different Industries so what happens quite frequently is those two industries speak different languages so it's important that that continuous communication is omnipresent and those people have open and honest conversation so that they can make sure that they're talking the same language otherwise frustration can build up quite quite quickly and sometimes that can lead to relationships to break down why is this important because fundamentally what we're seeking to gain access to is data and particularly like the FI the financial sector as with any other industry holds uh an awful lot of privately held data particularly relating to uh personal information uh so before we can open the doors so to speak to allow Academia into that world those relationships have to be built and the trust has to be present um and we can do that formally through uh you know security vetting and and non-disclosure agreements and so on but fundamentally th those type of relationships and that sort of trust is built on personal interaction so that has to be present as well uh but then the final ingredient that's necessary and wouldn't have been uh possible to produce this output uh is the access to data uh so what what these approaches cannot be built upon is Theory alone they have to be tested on real data and they have to be tted in real world environment so ultimately all of those proceeding steps a getting us to having access to that data uh because uh from an academic perspective and data science perspective once we get a hold of that we can work wonders with it but because it's s such personal uh information and Carries such high risk with it if it's h it handled in appropriately getting to that stage isn't straightforward definitely agree Eric I think uh Partnerships like these are uh pivotal in terms of driving Innovation and uh Bridging the gaps between the academic and the industry communities let me bounce back to uh Zubin so Zubin how do you view the role of collaborations in fostering advancements in Ai and improving compliance outcomes so I think when we look at it as again we think about it from the consumer perspective as a bank and we are thinking about this my problem statement I need to do it this way I have a hypothesis but then when you flip over and bounce the ideas off to a non- banker non fi they may look at it with a different lens and when they looking at it from a different lens they may come up with different sort of solutions which they may have looked at at wider researchers or Industries and that can be also brought in now what also comes in is from the world of compliance we don't deem it as a competitive Advantage as such from us it's like keeping the financial system safe so you have all the players uh involved in it it so if there is an innovation or something which happens at Unit A it's well worth for the others to look at it and try and deploy it at their place or even tweak it at their place to try and use it so that's where we think that more people having access to this sort of uh information from an academic uh research perspective helps the sort of little gears turn in their heads and possibly we get more input to strengthen the industry thank you so when your view on Partnerships underscores the importance in uh creating a real world impact uh Dr Eric to conclude how important do you think it is for Academia and the financial sector to align their efforts to address key challenges like money laundering using AI yeah I think uh it's essential and I don't see a world where uh the financial sector or any other industry can uh move forward without these sorts of relationships being more uh common and the reason for that is I think uh as many people have spoken about today globalization digitization increasing demand uh both Financial crime the Cyber Threat all of these issues are just going to develop into increasingly Wicked problems which take collaboration to address them if an industry or a particular organization tries to address these issues alone they will not succeed um so I think for that perspective to meet that challenge that's first and foremost why it's so important uh but also why it's important for them to work together is um having come from industry from a a law enforcement background um we Face similar challenges we we spend most of our day doing day-to-day business um you know keeping the flood floods at Bay um and that's what industry is focused on so they don't have time to lift their head above the parit and scan the horiz so Academia can do that for them if that open conversation exists and Industry puts forward problems Academia has the time and space to seek out Solutions and test them for them the other good thing that uh these relationships can do is um industry as well also has finite resources um so they don't often have the capacity and capability uh in human resources or intellectual uh resources and time and space to to look at these issues Academia is a free resource um you know as long as the the those objectives can be aligned Academia don't charge for these Services frequently U so they're there to be access free of charge any organization in theory can build this sort of relationship with a local academic establishment and move forward from there so as long as they can think of a problem and they can build a relationship to try and address that problem Academia will always be there to help offer a help in hand and why is that important to conclude I would just say it will help them meet the emerging demand because the line graphs on this are only going in One Direction it's increasing uh exponentially um so without coming up with more Innovative uh technological solutions that use AI that use machine learning that uh you know access to digital capabilities that are available to us particularly in this space um organizations won't be able to meet that and in this example you know we saved the equivalent of 45 full-time employees just by implementing this machine learning algorithm uh that's a huge amount you know from a human perspective but also from a financial perspective and that's not to say organizations should use these approaches to cut costs that's not what we would recommend but what it does enable them to do is to move those Human Resources elsewhere in the system uh to focus on other problems you know for instance in this scenario uh the suspicious activity reports the timeliness that that was achieved we managed to improve that by up to 2third uh so it was happening over 60% quicker than it was previously because so much human resources have been freed up and enabled that capacity to be focused where it was needed um so if you combine all of those things together I think they can work uh valuably together to to deal with these issues moving forward thank you Dr Eric your points on tackling complex challenges and enabling regulatory alignment through collaboration under underscores the importance of building strong and sustained connections between Academia and the financial industry ladies and gentlemen this research initiative is more than just a success story it demonstrates how leveraging advanced AI models can not only address pressing challenges like money laundering but also redefine operational efficiency and Regulatory Compliance I am pleased to announce that the full research paper developed by by the adgm academy Research Center and Emirates in partnership with Emirates NBD and radan Academy will be soon publicly available on our website This research project is a testament to what collaboration and Innovation can achieve and I invite you all to read it share it within your organizations and consider how you can adopt similar models to transform your own operations as we conclude I encourage you to continue exploring how AI can be a driver of positive change in your Fields together we can build a smarter safer and more resilient Financial ecosystem thank you for joining us [Music]