AI is in the press and public domain, with stories of machines thinking, of robots replacing humans and of machines taking jobs from people. Thinking machines are not new, Alan Turing predicted thinking machines as long ago as 1950.
Whether individuals or society fully recognise the impact there is now an underlying, and growing set of technologies, performing tasks that vary in degree of sophistication, from data and information processing through to ‘intelligent’ activities.
Over the last 2 years Generative AI (GenAI) and Large Language Models (LLMs) have come to the forefront of the AI push, ChatGPT and Bard to name a couple of popular applications. These provide answers and information in response to ‘user prompts’, in the language of humans. Importantly you do not need to be a coder to use them, see IBM for a definition of GenAI.
Within financial services firms, AI is being used, for both the customer-facing and backend processes. Firms have been re-skilling, bringing in data scientists, and building internal modelling, data, and knowledge capabilities. AI approaches lend themselves to an agile, experimental approach, and often can be applied to many business domains, problems, or opportunities within the firm. Whether at the front-end customer communication and servicing, or the many back-end (time-consuming) processes.
Financial crime prevention is a critical concern of any financial services firm, and often requires a substantial on-going investment and run costs. The core purpose can be summarised, as detect, prevent, and deter. However, anti-financial crime efforts come at a cost, both in terms of substantial systems and technology spending, and people costs.
Given the advance of AI, it seems a logical step to use it for the anti-financial crime ‘mission’ within the firm. This is now the time to understand and apply AI, taking advantage of the ‘agile’ approach to trialling, testing, and quick deployment.
Where to start? A good starting point is the business need and focus, what are the biggest issues firms need to overcome? What are the areas of focus, the threat landscape, the hot spots, the ‘dear CEO’ letters, and the highest cost areas?
In an ideal world, firms would be able to quickly and easily onboard customers, with limited ‘friction’, fully understanding the financial crime risk they present. On-going detection would be risk-focused, in real, or near time, with higher quality output without the ‘noise.’ Investigators would have an enhanced knowledge of the customer, customer transactions, and relationships – the ‘holistic’ view. Investigators would be ‘intelligent lead investigators’, applying human intuition and thought to decisions, as opposed to searching for information across multiple sources or clearing hundreds of ‘false’ positives. How could AI help firms move towards the goals of an intelligent lead, holistic approach?
Many firms have focused AI on processing, bringing data, and information together, to get to the point of human decisioning, faster with higher quality and accuracy. Techniques have moved the dial, Generative AI and LLM now provide a significant step change for investigations, co-pilot is a word you will hear in the investigations space.
Governance; What are your control and assurance processes across your business, data, and technology domains? Are individuals sufficiently empowered for this new domain, with the required knowledge and skills? Do they have sufficient authority to question and challenge, how developed is your model’s governance?
Approaches; How do the proposed approaches sit alongside current systems and processes? What will be the impact of applying AI to current solutions?
Business – where in the process are you focused, is it upfront on-boarding, detection, and prevention or is it the efficient processing of the outputs?
Techniques – which techniques will you use, for example, supervised learning to apply learned behaviour and best practice to investigations? Unsupervised learning to detect anomalies, or graph embeddings to identify hidden relationships? What about GenAI for co-pilots, supporting the human in their investigations?
Management and Data are key considerations, supporting the quality, accuracy, and ongoing use of AI.
In conclusion, AI is a wide topic with many techniques and approaches, however, the ‘science’ itself will not guarantee success. Firms should follow a framework, and then apply AI through trial, experimentation, and an agile approach to unlock the potential of AI in the fight against Financial Crime.
This editorial was first published in The Paypers' Next-Gen Technologies to Detect Fraud and Financial Crime Report 2024. The report explores how banks, fintechs, and PSPs are using AI and emerging technologies to detect and combat sophisticated fraud.
About Colin Whitmore
Colin Whitmore is an anti-money laundering innovation and technology subject matter expert, he brings over 20 years of experience within financial services, working with firms including Thompson Reuters, Aviva, Barclays, the Royal Bank of Scotland, HSBC, and Natwest. Mr. Whitmore advises senior leaders and decision-makers on business strategy and architecture including the use of processes, data, compliance technology, and innovation to improve anti-financial crime controls and meet strategic objectives.
About NatWest Group
NatWest Group is a relationship bank for a digital world. We champion potential; breaking down barriers and building financial confidence so the 19 million people, families, and businesses we serve in communities throughout the UK and Ireland can rebuild and thrive. If our customers succeed, so will we.
Every day we send out a free e-mail with the most important headlines of the last 24 hours.
Subscribe now