AI in financial services. Is the sector ready for the risks and rewards?

By Stuart Tarmy, Head of Industry Vertical Solutions, FinTech and Global Partnerships at Aerospike, explains why we need to understand and manage the inherent risks of AI.

AI is nothing new to the financial services sector. Numerous companies have been using it for years to gain competitive advantage, but that doesn’t mean the industry as a whole is truly reaping the full benefits. Indeed, different companies will choose to invest in AI at different points depending on their technology sophistication, resources, business use cases and leadership. An August 2023 Goldman Sachs survey of Fortune 500 CEOs found that they expect overall slow adoption at first with widespread adoption in the second half of the decade.

AI is being adopted across an almost endless list of financial services use cases, including fraud, analytics, customer 360, next best offer, fraud detection, anti-money laundering, risk, compliance, trading, liquidity management, credit scoring, document processing, sentiment analysis, customer support, call center automation and many more. On the surface, AI may be delivering the results required for a specific task, such as trading, but the technology is evolving at an astronomical pace, and that brings risk as well as reward.

New regulations are quickly emerging to address these risks. As recently as October 30, 2023, President Biden signed an Executive Order to regulate the use of AI in the U.S. and put in safeguards to manage risk and to avoid its abuse to the public detriment. Earlier regulations, such as the EU AI Act, which was originally proposed in 2021 and moved to the next stage of its enactment in June 2023, are seeking to ensure that broad AI controls are in place. The EU AI Act will place restrictions on how organisations use AI, for example as a facial recognition system or to manipulate behaviour. Where AI has been used to create content, AI-generated images or video that could be misinterpreted as ‘real’ (as we have seen with deepfakes) must be clearly marked. In addition, companies will be required to understand how their AI works, something that few know today or have had cause to consider in the past.

Given these developments, financial services companies need to start planning today how they’re going to manage the use of AI not only for their own businesses, but for the reputation of the sector as a whole.

AI’s role in financial services

Banking, payments providers, brokerage and insurance firms can all benefit from AI helping to improve risk and compliance controls over their activities, as well as to support trading decisions.

The checks involved in processes such as customer onboarding are a perfect candidate for AI, where a lot of manual work is still undertaken to meet Know Your Customer (KYC) and Anti-Money Laundering (AML) regulations. Risk and compliance checking is another area that will certainly see more AI automation going forward. We’ve seen examples of stock, bonds and derivative trades going through hundreds of compliance checks before the trade is approved to be executed. AI can help organisations better understand the risk, liquidity and financial impact of these trades at the individual customer, trade desk and corporate levels.

Lowering the risk of online payment fraud is another area that’s seen huge benefits from AI through deployments that enable real-time fraud prevention. One such example is where PayPal implemented a bespoke graph database as part of its AI stack. That solution has reduced its daily risk exposure by a factor of 30, while also improving revenue growth.

Understanding the ‘black box’ with Explainable AI

The financial services industry is already reaping rewards from the application of AI. But as discussed earlier, AI regulation is evolving across the globe, and this poses a risk for financial services and other industries. This is not because regulators want to stop institutions from using it, but they will require institutions to understand and demonstrate how their algorithms work and to have the checks and balances in place to ensure the AI does not do ‘bad things’, such as making biased or discriminatory decisions on credit card or mortgage applications.

Transparency is vital in financial services, where AI decisions must be understood and justified. Explainable AI (XAI) is the discipline that helps organisations ensure that AI decisions are transparent and traceable, essentially allowing them to see what’s happening inside the “black box.” It allows companies to provide faster, reasonable explanations to users, auditors and regulators to show that the system is not producing biased or discriminatory output.

One reason XAI is so important is because, what’s happening inside an AI system, is, by its very nature, self-programming and evolves every time new data becomes available – whether from an external source or as an analysis of the results from a decision made. It’s in a constant state of refinement and evolution, yet how it’s working needs to be clear at any moment in time.

But the move towards AI isn’t just about improving customer experiences and productivity. It also addresses the pressing need for financial institutions to effectively address certain events and long-tail scenarios, such as unanticipated shifts in interest rates, customer behaviour or global macroeconomic events. By harnessing the power of AI, they not only bolster their adaptability to unexpected situations, but also increase the likelihood that they’ll be in sync with regulatory expectations, all while staying ahead in the fiercely competitive financial landscape.

Real-time data is key to AI success

With the shift from traditional to modern FSI architectures, it’s important for the sector to embrace the multi-dimensional role of AI and understand the importance of integrating real-time data and transparency to fully realise AI’s benefits.

AI needs massive amounts of real-time data along with a robust infrastructure that ensures consistent performance at scale. The fusion of XAI and real-time data allows organisations to constantly improve their systems, and enables companies to move quickly while still meeting the needs of regulators and auditors.

XAI, combined with real-time data analysis, provides the vital capabilities of monitoring and understanding system responses and outputs in light of rapidly changing market conditions. It establishes a continuous feedback mechanism that equips financial institutions with the technology they need to proactively detect and manage emerging risks, ensuring they remain adaptable and competitive.

By embracing XAI and real-time data, financial sector companies can bridge technological complexity and practical use, thereby closing the gap between data scientists (who develop and fine-tune these algorithms) and business executives, who benefit from the ability to make better, more accurate decisions with the aid of AI, XAI and real-time data.

Control builds trust and confidence

The financial services industry is only just beginning to realise the benefits that AI can bring to customer experiences, operational efficiency, and risk management in all areas from payment providers to retail and corporate banking, capital markets, asset managers and brokerage firms.

As organisations explore the opportunities for AI, they must incorporate transparency and trust into their design. Their desire to use AI should not put these attributes at risk as they are the foundation on which the financial services sector is built. Combining XAI with real-time data will enable organisations to take a controlled, responsible approach to AI that will help them find their competitive advantage, build customer confidence and propel them ahead of their competitors.