Earlier this year, our CEO Julian Dixon was invited to chat to Claus Christensen, CEO and co-founder of Know Your Customer, as part of its regular RegTalks series. During the episode, Claus quizzed Julian on:
• The role of artificial intelligence in financial services
• How to strike the right balance between rule-based systems and AI for best results with transaction monitoring
• Regulators’ understanding and attitude towards the use of AI in the banking sector
In case you missed it or would like to recap, here’s a summary of 10 key points discussed:
1. The role of AI in financial crime
Regulators don’t require organisations to use AI, but the fact is, the future of all industries, not only financial institutions, is heavily dependent on technologies such as AI.
At Napier, AI has been in our plans from the outset, knowing that it would become increasingly relevant and important within this industry.
2. AI in transaction monitoring
The trend now is not to be asked “have you got AI”, but rather “if you have AI, does it work?”.
AI’s role in transaction monitoring is interesting because it is often assumed AI will replace traditional rule-based monitoring.
But this is not quite the case.
Regulators mandate that a risk-based approach and a robust, easy-to-use rule-based system has its place – but it must be effective.
Where rule-based systems aren't actually perhaps that effective, is where AI can be brought in. AI can find the unknown unknowns, or make the process more efficient and reduce workload.
3. Rule-based systems will always have their place – but will be enhanced by AI
A transaction monitoring system should reflect an organisation’s policies and procedures. And if the organisation has a risk policy that says: "Anything from country X over $1 million is suspicious”, then an analyst needs to look at it.
No organisation needs AI to tell you that. A rule will suffice. And so rule-based systems will always have their place. But over time they will be augmented and enhanced with AI.
4. Humans have an important role too
Despite all the benefits AI has to offer, you need human intervention on a regular basis. This goes back to the does it work question. Humans are needed to make sure AI stays tuned and doesn't stray from its mandate.
5. AI needs to be integrated into policies and procedures
AI always has to be somewhat aligned to the policies and procedures of the organisation. This, of course, is set at the C-suite level and reflects the risk appetite, within an organisation’s regulatory framework.
Going forward, it would be great to see policies and procedures starting to reflect technology in a more closely aligned way, with an understanding and recognition of what AI is and what it can bring to the table. The policy and procedure could say, for instance, “use adaptive profiling algorithms over this kind of data”.
6. Regulators are becoming receptive to AI
When it comes to the use of AI to fight financial crime, all regulators move at different speeds. Famously, MAS (the Monetary Authority of Singapore) is very advanced. And the US regulators also came out with legislation about 18 months ago saying you could start to use real historical data to test your AI models, and there would be no punitive action for retrospective things that the AI finds. This was very encouraging for financial organisations, allowing them to start testing AI technology in a real way.
In the UK, the FCA (Financial Conduct Authority) is setting up within its financial crime area a data science section. This will allow a greater understanding of how the technology works and therefore how it can be applied. And more importantly, it will allow the FCA to actually begin to understand what's going on when they do their audits.
The FCA takes a real big lead from the tier one banks in particular. The tier ones have typically had the larger budgets to have these bigger programs. The FCA is beginning to get a much greater knowledge and understanding, recognition and acceptance of AI.
7. Explainability can be a challenge for AI
Explainability has traditionally been a challenge. You can't justify decisions simply on the grounds of “the machine said that”. This is why machines and humans need to work together.
It’s also why our machines have got that explainability and a simple user interface experience. Explainability means the analyst can understand what the algorithms and machine learning have found, explain it in English and begin their investigations. When an audit takes place, everything is explainable.
8. AI applications are not always understood
One of the problems around AI, when we go and talk to customers, is they don't really know what they want the AI to do. People read widely about AI, so they can imagine what it can do and can't do, but when you actually ask them what they are trying to do, they’re not so sure.
9. Compliance and front office should both reap the benefits of AI
It's actually a really good for compliance and the front office to be sharing an AI-enhanced transaction monitoring system. We find account managers really love our technology because they can see the accounts in real-time (such as what the accounts are trading in, where they are focused on) and it really helps them manage the accounts as well. It really brings the front and back office together with the same piece of software in a very constructive and progressive way; business-enabling as well as compliance-enhancing.
10. Mandating regular technology updates would be revolutionary in improving the fight against financial crime
Following a final big question from Claus, Julian explained that if he was the global financial regulator, he would mandate regular technology updates: “In the UK, and I guess this is all over the world, you have to get your car tested to see if it's fit for purpose. I'm just using cars as a metaphor, because cars have to get more efficient over time. If they're not more efficient, they get taxed more.
“So I actually think it would be great that a global regulator has some kind of equivalent: a test and consequential tax on systems that are older and less efficient. Banks are therefore incentivised to keep their technology in pace with the global technology trends, and even more importantly, keep up with money launderers, who have no budgets or limitations to their talent.
“Money launderers are building tech every day of every week of every year to infiltrate the systems we have. Unless we use advanced technology, fighting financial crime becomes an ever more difficult war.”
You can listen to #5 – RegTalks with Julian Dixon here.
Discover our award winning AML technology
Request a demo to see our technology in action, or contact us to find out how we can help improve your AML processes.
.png)
