Reporting and Governance
Marshall Choy

Marshall Choy

AI is transforming risk management across all areas of financial services, demonstrating measurable risk impacts in credit, portfolio, market, fraud and underwriting, argues Marshall Choy, SambaNova senior vice-president of product.

Banks are the linchpin of our modern economy: influential, powerful, and tightly regulated. Given the banking sector’s fundamental influence in global economies and government policies, risk carries a very different meaning than it does in other industries. It is important to balance the needs of shareholders with the broader impacts of banks’ activities on the economy and the commensurate risk that carries, as we saw prior to the financial crisis.   

Banking operations involve thousands of documents, huge amounts of incongruous data, and repetitive, manual processing requiring significant human capital. With that in mind, we can see why current compliance processes are long, expensive, and prone to errors. So, to minimise their exposure to fines and penalties, banks increasingly rely on bringing the latest technology to bear, using risk management algorithms to manage the multitude of risks they face.

No system is perfect, but with the advent of artificial intelligence (AI), these algorithms can be enhanced to an unprecedented level, making them far more accurate and efficient. AI is poised to help banks maintain a healthy balance of risk in several ways: crawling dense contracts and documents, categorising them, redacting sensitive information and signposting potential risks before they become systemic.

But first, to understand where the industry’s going, let’s take a look at where it has come from.

Risk management is at the heart of banking

Risk management has always been front and centre for banks and risk itself is not a monolith. It appears in various forms, including conduct risk, credit risk, market risk, operational risk, reputational risk and liquidity risk. Each of these carries far-ranging consequences. On an individual level, if a borrower is unable to pay back money that a bank has lent them, this means increased collection costs and obstructed cash flows. Similarly, operational risks like information breaches, undertrained staff and technology failure threaten losses to the whole organisation’s bottom line.

With reputational risk, the consequences are just as serious. Through errors or individual lack of judgement, some banks’ decisions have prompted regulators like the Financial Conduct Authority to step in to demand that they pay restitution. This creates long-term issues like plummeting share price or market share, declining competitive advantage, and deteriorating customer relationships. In an extreme case, Natwest was fined £264.8m following its failure to comply with money-laundering regulations, which allowed the launderers to deposit £356m in cash at the bank. As a result, the bank’s share price declined by 4%, costing shareholders more than £1bn.

If banks do not adequately manage risk, their employees, their customers, and the society they operate in will all feel the sting.

Why banks currently use algorithms to reduce risk

By reducing the opportunities for human error, banks can de-risk their operations while expanding their offering to a wider and more diverse audience. Hence the need for strong automated algorithms and solid indicators to feed them.

Some of the most important algorithmic factors in assessing risk include leveraging, cost-to-income ratio and market capitalisation. These can be vital aspects of many risk analysis algorithms designed to reduce risk and over-leveraging.

By weighing these, and many other elements, into their calculations, banks can ensure that they balance ambition with healthy caution. They straddle the line between increasing profits while covering liabilities, using resources efficiently while generating profit, and increasing market share while protecting against sudden decline.

However, when it comes to systemic solutions, there is always room for improvement. AI can play a crucial role by helping banks to further optimise their risk management strategies beyond what is possible with the human eye alone.

How advanced AI models can sharpen de-risking algorithms

Financial services firms are turning to the next generation of AI and machine learning to develop highly accurate, augmented algorithms to analyse compliance and risk. Pre-trained models are ideally positioned to help them do this, using machine learning and a built-in understanding of financial terminology to pore over thousands of datasets and files with incredible speed. Increasingly, banks are employing all of these applications from a single AI foundation model: reducing sprawl and infrastructure spend.

AI can also proactively identify areas where banks are most at risk of regulatory backlash, and then take steps to mitigate those risks, such as using sentiment analysis on voice calls to evaluate customer complaints. AI could also help banks automatically redact sensitive information from contracts and other documents, saving time, money and sleepless nights.

The information unlocked by AI includes structured datasets like transaction information and customer references, semi-structured data like web logs, and unstructured data like chat records and voice calls. As a result, banks can integrate larger and more diverse datasets and process them quickly and accurately, resulting in an efficient and scalable analysis that legacy rules-based risk reporting systems cannot handle.

This has a series of concrete benefits: helping financial services leaders to better understand dynamic compliance requirements, integrating policies into business units, seeing risk-monitoring and moderation in real-time and locating new automatic trading triggers.

Risk management is at the very core of every bank, and thus every economy. With the use of algorithms as a long-standing tradition that allows employees to operate beyond their usual parameters, AI is the next step.

Bankers, by definition, must be comfortable working with uncertainty. Composure under pressure and an intuitive balance between ambition and caution set them apart from many other people. However, AI exists to augment these capabilities, not to replace them. With the latest advancements in AI, financial institutions can redefine their relationship with risk and act decisively, with increased confidence and greater speed.