Silent Eight Cosponsors IFCTF 2017 in Malaysia
Silent Eight co-sponsors the 9th International Conference on Financial Crime and Terrorism Financing (IFCTF) 2017 in Malaysia.
Organized by the Asian Institute of Finance (AIF), and supported by leading organizations including Al-Rajhi Bank, Bank Negara Malaysia, Securities Commission Malaysia, SIDC and the Malaysian Insurance Institute, the 2017 IFCTF was titled “Future Proofing Compliance: Responsibility and Response-Ability”. The event’s primary focus was on modern trends in financial crime and how future technologies will play become a key tool that financial institutions will utilize to meet regulatory legislation.
The organizers, AIF, aim to develop research on human capital development, and work closely with industry leaders to advocate for and implement professional and ethical standards.
Industry experts from around the globe provided their insights on threats and the strategies that are being developed to address these threats. With an estimated participation of over 500 delegates, the event has become one of the regional highlights within the anti-financial crime community.
During various presentations, the challenges associated with transaction monitoring were addressed by the speakers, with highlights including that artificial intelligence is the future of the industry and the key part of the process. While banks are looking to improve effectiveness and efficiency, they are also under regulatory pressure and they are continuously investing more resources in their transaction monitoring capabilities.
Mr. V. Maslamani, the Chief Compliance Officer of Al-Rajhi Bank, proudly announced that they will be the first to employ AI technology in their compliance process in Malaysia.
Martin Markiewicz, the CEO of Silent Eight, presented an introduction to Silent Eight transaction monitoring solutions powered by artificial intelligence. His presentation highlighted how AI can become an effective TM tool that works in harmony with other tools to achieve the intended outcome. To that end, it is important to clear any misconceptions related to what artificial intelligence can do for transaction monitoring. For example, If your only tool is a hammer, then everything around you starts looking like a nail. But in the real world, there is more to life than just nails – there are many other objects that need to be handled with tools other than the hammer, and it is important to have the right tool for the job.
Transaction Monitoring presents many different scenarios and depends on many different thresholds, and every case has to be investigated and filed. This is a manual process that is performed by analysts, and it is both time consuming and subject to human error. When we modify thresholds to catch new potential scenarios, we often end up with a disproportionately increased volume and frequency of alerts, making it virtually impossible to scale the system to address new threats.
No single AI technology can be a magical all in one solution that will fix all these problems right out of the box. The toolbox needs to contain a complex set of AI technologies. The good news is, all these technologies already exist and have been successfully applied. Many leading corporations like Uber, Google, Facebook, Amazon and Apple have developed specific AI based technologies to solve specific problems.
Siri is a personal assistant software from Apple. It can do a lot of things that a human assistant can do, like plan appointments and find venues for events.
But would you put Siri behind the steering wheel and ask her to drive you somewhere? That would be using the wrong tool for the job. However, self-driving AI does exist from other corporations, but it is not very good at making appointments for you.
Before AI based solutions are deployed, we need to teach them how to do their job. In our case, how to think like a transaction monitoring analyst. AI has a learning curve that starts at primitive levels, going up to human level and then eventually surpasses human capability in terms of effectiveness. For example, AI can learn how to play a game. Initially it will be sluggish and lose, then it will keep up with human contestants, and finally it will consistently defeat them.
Another example is the self-driving AI. This is initially trained via a simulator where there are no consequences for error before it goes on the road where there are consequences for errors.
In transaction monitoring, we can recognize three major steps in the process:
- Alerts being generated
- Analysts investigate the alerts
- Analysts come up with solutions
The AI powered transaction monitoring solution picks up the alerts that come in from transaction monitoring systems and tries to analyze them like a human analyst – which is exactly what it is trained to do. It retrieves data from many different places (like the web, news portals, as well as internal and external databases), analyzes the data very quickly and passes its solutions to a human analyst who then doesn’t need to spend a lot of time on investigating the alerts since they are already solved. The human analyst can either agree or disagree with what the AI has concluded. This allows us to scale the process as needed – we can modify the thresholds or settings and AI will simply carry the changed investigative workload in a scalable way.
A good example of a successful implementation of this philosophy is OCBC, which in partnership with Silent Eight, employs AI based TM solutions – with excellent results.
That explains just one way that we can apply existing AI technologies in TM.
Today it is not possible to replace human judgement capabilities by machine based technologies. That is not the aim of our AI based solutions. We have created a tool to empower human analysts, that works in coordination with other TM tools in order to improve both effectiveness and efficiency.