The rise of AI and machine learning in financial services is already driving major benefits across compliance and the customer experience. To realize the full potential, Sean Durkin, Head of Data Science at Barclays, tells us in our latest Expert Talk about the importance of being able to appreciate the “art of the possible”.
- Sean Durkin, Head of Data Science at Barclays, likens the potential of AI and machine learning in financial services to the dawn of online banking.
- AI can help to enable full regulatory compliance, minimize downtime, and promote quicker decision making, all of which will improve the overall customer experience.
- Machine learning in the form of a neural network is able to replicate the decision-making process, as well as take analysis a step further with an aggregate of multiple human decisions.
Best practice tells us that screening remains our best defense against corruption.
However, Refinitiv’s 2018 research into the true cost of financial crime highlights just how tough it is for often over-burdened compliance teams to pinpoint potential risk in the age of big data and global third-party networks.
Such challenges can lead to gaps in compliance, which inadvertently allow financial crime to flourish.
Our global survey questioned over 2,300 senior managers and C-suite executives across 19 countries and revealed that, despite respondents spending an average of 3.1 percent of annual turnover trying to prevent financial crime during the year preceding the survey, nearly half (47 percent) had fallen victim to corruption over the same period.
Leveraging AI and machine learning in real time can help to ensure regulatory compliance by enabling efficient, consistent and scientific decision making. It’s little wonder that the technology is making waves in financial services.
AI and machine learning benefits
In the latest Refinitiv Expert Talk, Sean Durkin, Head of Data Science at Barclays, likens the advent of artificial intelligence and machine learning to the dawn of online banking, in how much it will be “a paradigm shift for both financial services firms and their customers”.
He says the benefits for compliance teams are significant, especially when they face the familiar needle in the haystack scenario of pinpointing risk within a vast universe of available data.
- Machines can help identify entities and their relevance to content themes to reduce false positives.
- Intelligent tagging can structure content so that millions of text documents can be processed.
- People, places, facts and events can be tagged with scores to indicate relative importance.
Other applications include article de-duplication, which allows users to focus on unique events; and content grouping, which identifies and clusters similar content relating to a theme and a specific entity name.
But the value doesn’t end there. According to Durkin, one of the most significant benefits offered by AI and machine learning is that they can significantly help organizations to minimize any negative impact on customers.
AI can help enable full regulatory compliance, minimize downtime, and promote quicker decision making, all of which improve the overall client experience, which is a major focal point for most financial services firms.
Enhancing human decisions
A ground-breaking development in this field is the ability of machines to capture the decision-making process that the human mind undergoes in different scenarios.
Durkin explains that in financial services, one of the fundamental processes involves a subject-matter expert analyzing information (for example, a loan application) and making a decision.
Machine learning in the form of a neural network can help replicate the decision-making process, but also take the analysis a step further by producing an aggregate of multiple human decisions, effectively consolidating intellectual property and removing subjectivity and external factors.
Trust in new technology
Providing proof that these new data-driven technologies actually work is fraught with challenges, but Durkin explains a relatively simple method for building trust in the unknown.
By temporarily removing an algorithm’s access to data at a past point in time (for example, six months before the test date), the machine is allowed to predict a future that has already happened in the real world.
Comparing the predictions to what really happened measures the accuracy of the AI and machine learning techniques.
The potential of AI and machine learning
Durkin maintains that technology will be the key enabler in the next three to five years, commenting that innovations such as the cloud have unleashed enormous opportunity.
“The level of infrastructure you needed to do this work even 10 years ago was siloed within the limits of only the biggest tech companies.
“That’s where innovation and creativity was almost forced to take place in this space. Now it’s not. It’s federated and democratized and that’s a great thing for financial services and for data scientists.
“It’s a wider field because the ecosystem you have now within everyone’s grasp is extremely conducive to creativity, experimentation, and agility.”
Watch: What technology will transform the financial sector in the next 5-10 years?
Moreover, he believes that the true potential of AI and machine learning has yet to be discovered and will only emerge when we begin to appreciate the “art of the possible”.