Could generative AI be a game-changer for banks? (Part 2)

• Gen AI can help banks unlock value trapped in mountains of data – value that can also accrue to banks’ customers and investors
• By creating customised, personalised messaging, gen AI can deliver scalable, private banking-like services in a way that keeps clients engaged – and optimises revenue
• The expense of model training limits certain gen AI trial activity to the largest institutions, which may result in formidable “moats” that are difficult for smaller competitors to overcome

In our last post, we talked about how banks have traditionally been early adopters of technology that makes their operations more efficient. Yet this has also contributed to a technological “doom loop”. As banks compete to become more efficient, they offer their products and services at continually lower prices. But generative AI (gen AI) could be a game-changer. Here’s why.

Gen AI and banking: New opportunities, new benefits

First, it may be helpful to recap just what generative AI is. It’s not a single program or tool. Rather, it’s a category of artificial intelligence models that, based on user prompts, can produce original and coherent outputs, which includes text, images, music, videos, and even computer programming code. These generative AI models have undergone extensive training by digesting massive amount of data, allowing them to generate outputs that can mimic or extend human creativity, efficiency and problem-solving capabilities,

For banks, it’s the insights generated by AI that may prove to be most useful. Consider these many possible applications:

• Re-engineering processes. Gen AI can help banks understand the triggers for when a customer needs assistance and how to intercede. According to a study from Accenture, the banking industry could see up to a 30% improvement in productivity (i.e. potential hours saved) through the adoption of Gen AI.¹
• Identifying incremental revenue opportunities. Accenture also found that the average consumer has 6.3 financial products, but only half of these reside with their primary bank relationship.
• Enhancing customer experience and driving deeper engagement. We see gen AI delivering scalable, private banking-like services in a way that keeps clients engaged – and optimises revenue. Advanced recommendation engines built with large language models (LLMs) will be capable of creating personalised messaging that considers a customers’ unique circumstances and goals.
• Generating efficiencies. Consider the person-hours saved by using sophisticated, gen AI-powered chatbots, or by employing “copilots” that can help streamline the processing of mortgage documents.

Gen AI is not without its challenges

Despite all the potential benefits, gen AI adoption rates vary widely among banks and other major financial institutions. This may be because gen AI has its own notable challenges:

• High costs. The complexity and expense of model training is limiting trial activity to the largest institutions. Over the long run, we expect costs to come down, enabling broader adoption. At the same time, large banks’ ongoing investment may result in formidable “moats” that are difficult for competitors to overcome.
• A spaghetti infrastructure of complex systems. Many banks grapple with how to incorporate AI with their built-for-purpose hardware and legacy programing languages. Fortunately, Gen AI itself could hold the key to contemporising the core of banks’ systems through reverse-engineering legacy code and translating it to modern programming languages. Goldman Sachs says gen AI tools have helped it rewrite 40% of its legacy code base.²
• Regulatory concerns. Banks don’t want to get too far ahead of regulators who may not yet be comfortable with gen AI systems. They need to build their models with explainability at the forefront, and must closely engage with regulatory bodies.
• Self-imposed caution. Many banks are well aware of how automated workflows can help them pare down or even eliminate some simpler functional roles. However, some executives are cautious of how this might hurt their talent development runways and overall organisational structures.

The bottom line

Banks are no strangers to technological change and disruption, and they have a long tradition of investing heavily to keep pace with their peers and emergent fin-techs. While this has helped reduce some costs, banks have seen little benefit in their cost-to-income ratios. As certain costs have fallen, regulatory burdens have grown, and it has become more expensive to attract and retain customers.

That’s where we believe gen AI can help. It certainly has the ability to streamline operations – including client service, marketing, compliance and other shared service/cost centres. But more profound, in our view, is the opportunity for gen AI to reverse some of the detrimental impacts of prior technologies (including earlier AI systems) that have made more financial products available at lower costs, but at unclear value for banks or their customers. We believe Gen AI has the potential to help unlock value trapped in siloed systems and data – value that can accrue to banks, their customers and their investors.

¹ Source: Accenture, Jan. 2024, “Top 10 banking trends for 2024”
² Source: CIO Dive, Jan. 2024, “Cracking COBOL: Banks to deploy AI to retool legacy apps”

Back to Blog