As artificial intelligence (AI) takes on a growing role in financial services, from granting loans, to automating small-business bookkeeping, banks grapple with how to use it responsibly and authorities seek to put rules in place.
Ciko Thomas, group managing executive of personal and private banking at Nedbank, says the ethical use of AI is “a big factor in banking. Already, we're grappling [with this] a lot as an industry, and I know we've done a lot of work inside Nedbank to think about the ramifications of AI in our business.”
Thomas says Nedbank is interrogating how AI intersects with risk management, fraud prevention, anti-money-laundering and customer experience. “We can use that data as a massive force for good… or we can choose to use it for dark reasons.”
Banks, he adds, are “sufficiently well regulated” and have “sufficient safeguards” in place to ensure ethical AI use. But he warns that “other parts of the financial services value chain… are not subject to the same level of scrutiny”.
This is why Nedbank actively works with partners, including fintech firms and service providers, to ensure they also follow strict rules, notes Thomas. “As an industry, we are constantly looking to make sure that, even as we work with partners, whether it's other fintechs or service providers, or suppliers, we become guardians of those ethical standards. And it's an ongoing conversation.”
Thomas believes there is not enough public debate on these issues. “I don't think that at country level, we are having these conversations enough around the ethics and value judgements.”
The South African Reserve Bank’s Prudential Authority (PA) is only now formally examining AI and related concerns, such as cloud computing and data offshoring. It has issued interim guidance to banks and insurers calling for a risk- and principle-based approach, while it develops a policy framework.
The PA is running a survey on AI adoption in banking, with results expected later this year. The study will address safety, soundness, consumer protection and financial stability.
Cambridge University Press last year described South Africa’s digital sovereignty as “at the crossroad of securitisation and development”. It noted that government entities “are struggling to cope with increasing responsibilities… to protect citizens’ rights to data protection and safety and security online”.
Mark Walker, MD of IDC South Africa, previously said data sovereignty has become a significant concern across the industry.
Balancing innovation and trust
Thomas says banks are not “pushing the boundaries of what AI can do”. He views this as a positive, as “it keeps us honest in terms of how we think about how we can unlock the power of these tools when it comes to ethical values and making sure you are absolutely committed and bound to trust”.
AI is a “powerful tool” for financial services, says Thomas. “If we look at AI, can it be a catalyst for SME growth in South Africa? Without a doubt it can.”
He adds that cutting-edge tools help democratise lending because algorithms can be used to assess customers’ financial status.
Fintech players are already exploring ways to use AI and real-time data to democratise payments, reduce cash use in spaza shops and optimise stock orders, he says.
Lungisa Matshoba, Yoco co-founder and chief product officer, says there is “a large amount of interest in AI and the role it can play in helping small businesses with streamlining the process to get insights right down to the smallest details”.
AI, he notes, could automate stock analysis, speed up bookkeeping and give owners deeper insights. “This is something that really excites us. We want to support business owners even more.
“It must benefit our customers… What you have to look at is how do you leverage AI to create real insight that is unique and distinct to that specific business.”
Matshoba says some small business owners are “genuinely concerned and scared” about AI’s impact. “I'm hoping we start to see more business owners in this conversation… so they can start to leverage AI.”
Thomas stresses that human oversight remains key. “With these large reasoning models, human reasoning must always be applied in how we use these tools.”
Yoco is still in the exploratory phase. “We're not there yet to launch or release anything… but this is a good direction for us to enter,” says Matshoba. Yoco also, with permission, makes use of information captured through its point-of-sale devices to enable cash advances.
Share