By Hani Hagras, – Chief Science Officer at Temenos
The use of Artificial Intelligence (AI) in the financial services sector has grown exponentially in the last decade. Banks are now implementing the technology across a range of innovative use-cases. A new survey of IT executives in banking, conducted by The Economist Intelligence Unit (The EIU) in partnership with Temenos, finds that 85 per cent have a “clear strategy” for adopting AI to develop new products and services. Heavy investments by banks into AI also demonstrate this upward curve, with JP Morgan Chase, for example, spending US$12bn a year on technologies such as AI and machine learning.
The last few years have seen significant structural changes in the banking space. It has been driven by key pressures from the climbing cost of services, increasing competition from Big Tech and start-ups, and growing consumer pressure for seamless customer service. In the past, AI has primarily been used to automate routine tasks. But banks are now seeing it as a vital tool to overcome these growing pressures, fuel back-office efficiency gains, support product innovation and develop new business models.
With the data-driven nature of the banking industry providing fertile soil for artificial intelligence – paired with the increasing recognition of AI’s breadth of application – it comes as no surprise that AI is becoming a game-changer for banks. According to a separate global survey of senior banking executives, four in five agree that unlocking value from AI will distinguish the winners from losers in the industry.
As innovation in the banking sector grows, the list of functions of AI technology symbiotically expands. The EIU survey provides insights into AI’s key uses for banking operations. Fraud detection is the top application of AI by banks. By collecting data on transactions and authorisations, banks payments services provider Mastercard uses AI to predict and detect fraud more precisely and quickly. Improvements in data accuracy not only reduce financial losses but also false positive transactions and the hold-up of legitimate transactions, helping to improve customer experience.
Customer demands for a seamless banking experience are becoming the gold standard. AI offers numerous uses to provide at-scale personalisation to anticipate customer needs and create highly tailored services through the use of individuals data. This includes deploying AI within the back office to optimise and streamline IT operations. And improving customer support through chatbots and innovations such as “smile-to-pay” identification to ensure frictionless transactions. Other applications include embedding AI within Buy-Now-Pay-Later (BNPL) services to offer a personalised service. This encourages responsible spending practices, tracking consumer behaviour to support targeted advertising. While also using machine learning techniques to analyse customer transactions in real-time to accurately calculate default risks and offer cheaper loans.
However, while AI provides a breadth of opportunity, banks understand its limitations. As found in the EIU report, 62 per cent of banks agree that the complexity and risks associated with handling personal data for AI projects often outweigh the benefits to customer experience.
Most notably, trust and bias continue to be prominent barriers. For example, Apple experienced an unfortunate AI-related incident in 2019. Its algorithms – used to decide whether to grant credit lines – ran into claims that it was exhibiting gender bias, allocating relatively fewer credit lines to women than men. This is just one example of the adverse consequences of AI-based decisions. Thus, highlighting the need for transparency to restore trust from consumers and regulators alike.
With these risks at the forefront of AI technology for “high-risk” applications, such as credit scoring, banks are likely to feel increased pressure from regulatory and customer influences. Specifically, the need “ “explainability”, highlighting how an AI system makes decisions. Banks will need to establish a set of processes that allow users to understand the output created by machine learning algorithms. The success of AI will be enhanced with the use of explainable AI to spot and correct potential flaws and vulnerabilities in models.
Looking at this through the specific example of BNPL, Explainable AI (XAI) can explain to customers why a particular BNPL package was recommended to them. This provides customers with the transparency and information to take control of their own banking decisions, increasing their trust and understanding of the service.
The inclusions of strong explanatory capabilities will continue to be a core requirement for AI technology to gain customers’ trust, while also ensuring banks meet the likely tightening regulatory requirements. EU regulators have already announced they are seeking to establish stricter rules around the use of artificial intelligence in areas like crime prediction, credit scoring, employee performance management and border control systems. In particular, the bloc is seeking to mitigate undesirable outcomes and risks arising from AI-generated decisions.
While AI is clearly a game-changer, banks will need to establish a holistic strategy to ensure it is safe and suitable to help improve security and trust. And by ensuring any potential biases are mitigated, society can benefit from the market opportunities that AI has to offer in the fast-changing banking environment. As banks continue to innovate to meet and exceed consumer demands, they will need to tread carefully to ensure risks are balanced and services are future-proofed to handle the potential of AI.