How South African banks can deploy and benefit from Generative AI.

Say goodbye to manual processes and hello to improved efficiency and innovation. With Generative AI, SA banks can enhance customer experiences and fraud detection, and streamline operations.

Executive summary

• Advanced algorithms analyse vast amounts of data, providing personalised insights for customers, ensuring tailored financial solutions.
• Heightened security measures help combat fraud, protecting customers' hard-earned money
• Automation and optimisation drive cost savings and faster transactions, empowering businesses and individuals alike.

In EY’s most recent South African banking benchmarking report, South African banks have showed resiliency and reported some of their strongest interim results despite challenges such as increased geopolitical tensions, floods in KwaZulu-Natal, renewed electricity disruptions and rising inflation.

We found in the report that amidst this, banks remain focussed on three strategic themes: Cost containment and ongoing improving in cost to income ratios; Improving customer experience to drive customer acquisition and retention; and increased pace of digitisation, with significant growth in digital transactions and digitally active customers across all three banks.

Now, Generative AI is a ground-breaking technology with the potential to revolutionise the banking industry in numerous ways. With its ability to analyse vast amounts of data and generate new insights, generative AI is becoming an increasingly powerful tool for banks to enhance their operations and improve customer service. We examine the potential for generative AI to be a part of bank’s approach to realising these strategic intentions.
 

Cost containment and on-going improvement of cost-to-income ratios

Fraud prevention is one area where generative AI can have a significant impact on banking. By analysing transactional data and identifying patterns that may indicate fraudulent activity, banks can utilize generative AI to create predictive models that alert them to potential fraud before it occurs. This proactive approach helps banks prevent financial losses and safeguard their customers' accounts from unauthorized access, fostering trust and security in the banking ecosystem.

Generative AI now opens up more spaces for automation. Generative AI will take this to the next level where banks are able to automate even more of their routine administrative tasks with the technology. This may include enhancing current automation technologies, for example in the area of digitising documents, for example paper based loan applications or processing vendor invoices; combining OCR and language AI the system is more accurate as it can compensate for spelling errors or poor scans through AI but also able to generalise and deal with invoices that change in format etc. without human intervention. So, where the invoice number moves from the top left to top right for example due to the vendor updating their document template – as anyone who has implemented document intelligence solutions will know.

However, the major opportunity is certainly going to be enhancing the productivity of knowledge workers through the use of co-pilots. Examples here are too many to list but consider the number of management reports, PowerPoint presentations and Excel models that exist in modern banks – all developed manually by humans. Using generative AI technologies to pull data off the system and generate your board report or response to a customer complaint will mean a productivity gain of 2-3x (find reference for productivity gains). Humans will need to be kept in the loop as the accuracy will need to be checked for now.

One surprising benefit of generative AI in banking is its potential to enhance credit scoring models. Traditional credit scoring methods rely on historical financial data, such as credit history and income, to determine an individual's creditworthiness. However, generative AI can go beyond these limited factors and incorporate a broader range of data points. By analysing unconventional variables such as social media behaviour, online shopping patterns, and even sensor data from wearable devices, generative AI can provide a more comprehensive and accurate assessment of an individual's credit-worthiness. This approach has the potential to open up access to credit for individuals who may not have a traditional credit history but demonstrate responsible financial behaviour through alternative means. Another intriguing application is the creation of virtual financial advisors.

Improving customer experiences to drive acquisition and retention

Improving customer service is another major advantage of generative AI in banking. Through the implementation of generative AI-powered chatbots, banks can provide round-the-clock support to customers, addressing queries and resolving issues in real-time. These chatbots can be programmed to understand and respond to customer needs, delivering personalised recommendations for financial products and services based on individual preferences. The result is enhanced customer satisfaction and engagement, as well as improved efficiency in query resolution.

Co-pilots s as described above can also help in the front office but the systems need to be imbued with company specific data. When EY has applied this to the customer facing staff, the benefits will include being able to respond to the customer query much more quickly (less time on hold), more accurately (these systems can reference internal documents) and more consistently, and with less reliance on on-boarding and training which means that staff turnover is less of a challenge.

Customer experience with the tied agent force improved with more prompt responses to questions, less time on hold, fewer transfers between departments (especially where the query was a simple one).

Enhancing customer experience through a new interface

Chat bots will likely see another wave of use across financial services to handle customer service requests much more efficiently. It is up to the industry on how well it designs these experiences as to whether it will improve or detract from customer experience – in other markets such as the US there is no shortage of examples of frustrating implementations that annoy customers.

However, this is the tip of the iceberg. The big opportunity which we see in retail, for example Instagram, are a new type of interface that relies on natural language. On the new Instagram app, rather than a traditional search for “vegetables”, a customer can type “I want to make my kids a healthy meal for dinner” and the semantic search will yield a selection of appropriate items that go far beyond vegetables.

Taken into a banking context, we may see apps in future where customers are able to use natural language to interact for example “pay my mother” rather than navigating a maze of menu items.

Generative AI is not a panacea and risks should be considered

However, along with its benefits, generative AI also presents challenges that must be addressed to ensure responsible and ethical usage. One significant concern is the potential for algorithmic bias. If the algorithms used to generate insights are not designed with diversity and inclusion in mind, they may inadvertently perpetuate discriminatory practices. Banks must prioritise fairness and transparency in algorithm development to mitigate this risk.

Data privacy and security are additional critical considerations when implementing generative AI in banking. Banks must take measures to ensure that customer data remains confidential and secure, adhering to privacy regulations and industry best practices. It is essential to strike a balance between leveraging data for valuable insights and safeguarding sensitive information.

Open AI and IBM were in front of the US Congress this week in an open discussion on the need to regulate artificial intelligence both at a national and global level. What is clear from that discussion was it is just a matter of time before artificial intelligence regulation in the US is implemented. EU has a draft bill that is going to be promulgated in the near future. It is therefore quite likely that regulation across other nations like South Africa will follow suite quite quickly.

That being said, Generative AI also plays a vital role in regulatory compliance within the banking sector. The technology can assist banks in analysing large volumes of transactional data, identifying suspicious activities, and ensuring adherence to regulations. By automating compliance processes, generative AI helps banks mitigate risks, avoid penalties, and maintain the integrity of the financial system.

In conclusion, generative AI holds immense potential to transform the banking industry not just through shiny proof of concepts but as part banks’ broader business strategy. However, banks must navigate challenges related to algorithmic bias and data privacy while ensuring responsible implementation.