The media is currently awash with articles and comments on Artificial Intelligence (AI), primarily fuelled by the emergence and ongoing development of OpenAI's ChatGPT - an all-encompassing chatbot that has captured the public's attention with its impressive capabilities. ChatGPT is an example of a new generation of Generative AI tools that use machine learning techniques to generate brand new, 'original' content, such as text or images.

The recent rapid advancement in AI is anticipated to persist, and even further accelerate, reshaping industries through its transformative potential, while also presenting new risks and challenges for society. More recently, the EU has reached a provisional political agreement around a new AI Act, in an attempt to regulate this emerging technology.

Before the emergence of ChatGPT, machine learning (ML) was already an established field that has been widely applied in various industries. In financial services, ML applications can process and analyse vast amounts of data, such as transaction records, customer profiles, and market trends to identify patterns and insights that inform better business decisions.

In this article, we will focus on the application of ML in the UK's financial services sector, referring to the 'machine learning in UK financial services' survey conducted by the Bank of England (BoE) and Financial Conduct Authority (FCA), published in October 2022. In our two subsequent articles, we will discuss the risks of ML and how these risks can be managed effectively.

As we are all, at heart, forensic data analysts who use data and technology day-to-day to investigate legal and regulatory issues, our interest in ML naturally gravitates towards two main topics:

  • How can we use ML as a tool to improve the efficiency and effectiveness of our investigations of the digital fact base in a case?
  • What are the implications when ML models do not behave as intended and therefore become the focus of the investigation itself?

We will explore these two topics in this series of articles and will share our experiences on the benefits of using ML and other advanced analytical techniques in legal disputes and regulatory investigations.

While the terms AI and ML are often used interchangeably, they do not refer to the same concepts or processes. AI is a broad term encompassing multiple disciplines and techniques, with the primary goal of developing computer systems that can solve problems by mimicking or even improving on human intelligence. ML is a subset of AI that uses mathematical models of data to enable computers to learn without explicit instructions. Although these definitions could be further expanded and debated, they are sufficient for the purpose of this article.

The machine learning in UK financial services survey

In October 2022, the BoE and FCA published a summary of their latest survey on the current state of ML applications in UK financial services. The survey collected responses from a total of 71 financial institutions (FIs), including banks and insurers, to understand the use cases, maturity, benefits, and risks of ML applications. Some of the key points that stand out to us are:

  • ML is no longer just an abstract and theoretical solution searching for real-world problems to solve. It is now widely used in the UK financial services sector, with 72% of respondents stating that they are using or developing ML applications.
  • The use of ML is expected to grow rapidly over the coming years, with respondents predicting a 3.5 times increase in use within the next three years.
  • ML has moved beyond niche roles or proof-of-concept stages, with 79% of respondents saying that they have ML solutions in the latter stages of development and deployment.

The widespread use of ML in the UK's financial services sector means that it has truly arrived, bringing with it benefits and risks. While the increasing prevalence and benefits of ML applications are evident, it is important to recognise that the potential risks that come with their successes are complex, nuanced, and often understated.

The benefits of machine learning

The use of ML in UK financial services offers several benefits for consumers, FIs, and the financial system. For consumers, ML can provide more personalised products and services, as well as improved customer engagement and outcomes. For FIs, it can enhance data analytics and increase operational efficiencies. ML can also be used by FIs to improve the effectiveness of compliance systems for detecting and deterring unwanted or criminal behaviour, such as money laundering or fraud. These benefits can have an aggregated positive impact on the financial system and the economy as a whole.

The BoE and FCA survey suggests that all benefits of using ML are expected to increase in the future, with enhanced data and analytics capabilities, increased operational efficiency, and improved combatting of fraud and money laundering being the areas that are most likely to benefit from ML applications.

Our experience of using machine learning

Our forensic data analytics team regularly leverages ML techniques when undertaking large-scale document reviews for investigations and document production for disclosures. These ML techniques have significantly reduced the time and effort required to process large quantities of documents and assess their relevance to a particular matter.

For example, our recent work on the Post Office Horizon IT Inquiry highlighted the power of ML when faced with the daunting task of extracting insights from a large, unstructured, human-generated digital fact base. In this case, it involved approximately 60,000 IT tickets for logging and remediating errors on an IT system dating back to the late 1990s. We used non-supervised clustering models to identify the key themes within the corpus of IT tickets, and supervised models to identify tickets that are related to similar issues in the system. Our use of ML was instrumental in facilitating the investigation and enabling us to produce the expert report in three months.

Conclusion

In summary, the survey revealed that the use of ML in the UK financial services sector has grown significantly, and this trend is expected to continue in the coming years. As ML applications become more mature, they offer a wide range of benefits to FIs, including better customer insight and engagement, improved regulatory compliance, and increased operational efficiencies. As ML continues to evolve, it will undoubtedly play an increasingly key role in the financial services sector and beyond. However, it is crucial to acknowledge the risks associated with deploying ML applications in financial services.

In the second instalment of this series, we will delve into the ML risks in financial services, and our third article will discuss how these risks can be effectively managed.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.