Banks ramp up spending on artificial intelligence to boost efficiency

7 Jul 2024

Banking institutions have been big spenders on tech and they are seeing the results of employing artificial intelligence not only to improve their clients’ experiences, but their staff’s work too.

Banks globally rank among the largest tech spenders outside the tech industry, with some large banks allocating close to 20% of their expenses to technology, according to a Citigroup report on artificial intelligence (AI) in finance.

In South Africa, banks have quietly been raising their spend on technology while increasingly deploying AI to make the banking experience more seamless. Typical uses include facial recognition software – in other words, the use of a “selfie” to sign into your profile or even to open an account.

Jörg Fischer, Standard Bank Group’s chief information officer, says generative AI technology is still new, but he acknowledges it is constantly evolving and maturing. “We are working with global technology partners to provide all our people with both fundamental and specialised generative AI training and accurate information, and demystifying any uncertainties,” he says.

Fischer says Standard Bank is focused on productivity gains – working faster and more efficiently by improving software engineering and coding practices (helping engineers to fix bugs and codes) and really improving digital labour, contact centres, sales, service and administrative work.

“Our AI approach has moved from an ‘or’ to an ‘and’ one. This approach signifies the integration of AI tools in an interchangeable manner, leveraging their complementary capabilities without placing a singular bet on any technology partner owing to the speed and advancement of the technology,” he says.

The bank has also introduced coding assistants to help developers to generate code from AI services more efficiently. “Moreover, the automation of certain tasks has made it possible to scale operations beyond what could be achieved through manual efforts alone,” Fischer says.

At Standard Bank, AI tools such as Microsoft’s GitHub Copilot, Amazon Web Services’ CodeWhisperer and IBM’s Watsonx are boosting software engineering efficiency. Several users are testing Microsoft Copilot to unlock the power of AI in everyday tasks, while tools across various platforms like Salesforce and Microsoft Azure are being used to enhance productivity and streamline workflows.

FNB was an early adopter, having initiated large-scale use of AI about a decade ago, with a number of key use cases. It rolled out online fingerprint verification in 2014, allowing it to verify customer identities more securely and efficiently.

Use of AI for customer experience

On the customer-facing side, Standard Bank is using generative AI such as embedded OpenAI models in Salesforce Einstein to create pre-generated communications, gauge customer sentiment, employ AI-powered chatbots and provide predictive insights.

The bank uses what is known as software as a service, a software distribution model in which a cloud provider hosts applications and makes them available to end users over the internet.

“As these software providers mature with these AI technologies, we will take advantage of the embedded AI that will be available in their software updates,” Fischer says.

Blessing Mgaga, a divisional executive of client experience delivery at Capitec Bank, says the bank uses neural networks to analyse client transactions to personalise services based on patterns such as salary frequencies. “We’ve also integrated generative AI into our conversational banking platforms like WhatsApp, improving response accuracy and client satisfaction,” he says.

“FNB developed analytical AI models to combine information across a multitude of systems and sources for the purpose of understanding customer money laundering and terrorism financing risk,” says Christoph Nieuwoudt, FNB’s chief data and analytics officer. “We also introduced facial recognition technology on the banking app in November 2017.”

He points out that this feature enhanced the security and convenience of customer authentication by also using biometric verification against Department of Home Affairs records.

FNB started large-scale use of generative AI models in 2019, which combine the ability to detect and understand financial crime patterns and anomalies with the ability to generate detailed written reports to the Financial Intelligence Centre. Whereas initial implementations used internal, small language models, later approaches now also use trained large language models in conjunction with internal models.

The newer banks on the block, such as Discovery, Bank Zero and TymeBank, have the advantage that AI has been incorporated in their systems right from the start. Stuart Emslie, the head of analytics and data science at Discovery Bank, says AI has been embedded in the philosophy of the bank since it was launched in 2019.

“AI has now been embedded throughout the organisation, incorporating modern data-driven methods for product pricing, business monitoring, risk mitigation and shared-value creation,” Emslie says.

Using the Databricks Data Intelligence platform, Discovery Bank is combining data and actuarial science with behavioural economics, AI and machine learning to create data products and hyper-personalised experiences that reward healthy banking and lower financial risk.

Improving staff efficiency

Mgaga says leveraging generative AI in Capitec’s conversational banking service team has significantly enhanced staff productivity, particularly in client service operations over WhatsApp.

“Since the implementation of AI, we’ve observed a remarkable 2.1 times increase in efficiency. This improvement is measured through faster response times and our enhanced ability to handle greater volumes of client inquiries simultaneously,” he says.

Nieuwoudt says AI is showing solid returns at FNB, freeing up employees to be more efficient. “In the last financial year, more than 160,000 investigations were processed using the AI system. It has transformed the way that the risk assessment is conducted in the bank, because people are focused on decisioning – they’re not focused on collecting data and writing the reports.

“This has resulted in reviewing profiles for various financial crimes holistically, more rapidly and more consistently, and freeing up, on average, at least 70% of analysts’ time,” he said.

FNB has deployed an internal AI agent for staff, as well as an internally developed “code assist” agent for developers and data and analytical staff.

“Over the medium term, we expect the bulk of staff to use generative AI daily and for it to have a material impact on productivity and efficiency. Similar efficiencies are being realised where AI models are being leveraged to automate the verification of ID and passport documents,” Niewoudt says. DM

This story first appeared in our weekly Daily Maverick 168 newspaper, which is available countrywide for R35.

You may write a letter to the DM168 editor at [email protected] sharing your views on this story. Letters will be curated, edited and considered for publication in our weekly newspaper on our readers' views page.

Categories:

Tags:

DM is much better if you are signed in...

Business Maverick

Use of AI for customer experience

Improving staff efficiency