Banking on (Artificial) Intelligence

Adoption of Artificial Intelligence (AI) in financial services is expected to drive significant growth and productivity in the industry. Though some of the hype around Generative AI (Gen AI) is cooling down, the business opportunities are real. According to research conducted by PWC, AI will add a 14% increase to global GDP by 2030, equivalent to a growth of $15.7 trillion! AI adoption has risen rapidly driven by early adopters. Based on the Global AI adoption Index published by IBM, about 42% of enterprise-scale companies (> 1,000 employees) report having actively deployed AI in their business.

Gen AI is being applied across several areas of financial services. The top 3 use cases of Gen AI are in customer service, risk management, and software development. Large Language Models (LLMs) can perform tasks such as summarisation, content generation, classification, semantic search, code generation, and extraction, with estimates suggesting a 40% productivity gain in many of these areas. IBM has built AI Virtual Agents for 7 of the top 10 UK banks, that support over 30 million customer chats per annum. Over the last 2 years, we have used Gen AI in these virtual agents to improve customer experience and provide a better service. For example, REDI a virtual agent created by IBM consulting in partnership with Microsoft has a higher Net Promoter Score (NPS) than human agents by 39%.

Generative AI can have a big impact on fraud detection. By analysing massive amounts of transaction data, AI can identify unusual activity and flag potential fraud before it becomes a bigger problem. Language models (LLMs) are especially good at working with text data, which means they can help financial institutions analyse customer feedback, review documents quickly, and protect sensitive information.

Banks and Insurance companies have several old legacy core transaction processing platforms. Many of these platforms were developed 30-40 years back in archaic programming languages that a very few people have skills in. IBM is successfully using Gen AI to reverse engineer and modernise these legacy platforms. For a building society in UK, we are modernising the core mortgage platform. Gen AI is helping our software developers by reading the code of the mortgage platform and generating business logic, translating the code to modern languages and testing the application. In many cases, this can reduce the time and cost of modernising legacy platforms by over 50%, thus improving the ROI on such programs.

While AI has the potential to transform customer interactions, and business operations in financial services, adoption of this technology comes with its challenges. Given the industry is highly regulated the question of how to safely exploit AI is as important as where you are going to apply it. There are several considerations for building trusted AI, including data privacy, IP, transparency and explainability, compute and carbon cost, skills scarcity and, most important of all, governance.

Strong governance is central to building trusted AI, especially in financial services. It is crucially important to understand what AI models the organisation has, the data that the organisation tunes and applies those models to, the models’ intended uses, and their compliance with regulations. At least five countries have AI regulations, and two-thirds of the world's countries have privacy and data governance laws. Promontory (an IBM subsidiary specialising in regulations) is supporting many banks and insurance companies in the UK to develop strong AI governance and compliance to regulations.

Generative AI is all about people, so the industry needs to invest in training people on AI. Putting this technology into the hands of users, across all functions and lines of business—not just technology users—allows everyone to understand how transformative it can be to their role and the workflows around them. IBM is driving AI-first skilling across its enterprise, intending to train 2 million learners in AI by the end of 2026. We are also making AI skills content available externally via the Coursera platform. As AI continues to transform the financial services industry, the challenge lies in upskilling people many of whose jobs will dramatically change.