In South Africa, banking exists in two worlds: the marble floors of Sandton and the bustling streets of Mamelodi or QwaQwa. One world runs on algorithms and payslips; the other thrives on trust, hustle and informal cash exchanges. The idea that artificial intelligence (AI) could bridge these two realities once felt far-fetched, but that future is already here.
AI is quietly transforming how banks and fintechs approach inclusion. It can now assess creditworthiness using alternative data such as airtime recharges, electricity tokens, mobile-money activity and even social-behavioural indicators. This enables people without payslips or formal credit histories to access microloans, savings accounts and insurance. AI also supports automated onboarding, biometric verification and real-time fraud detection, all from a low-cost smartphone.
Yet, there is a catch: AI can only serve those it can “see”. Millions of South Africans still operate outside traditional data systems, especially in the informal economy. A taxi driver’s daily cash earnings, a vendor’s R500 weekly profit from amagwinya sales or a stokvel’s monthly rotation of funds; these are legitimate financial activities, but they rarely appear in the datasets used to train banking algorithms. When AI models are built only on conventional financial data, they risk entrenching the very exclusions they aim to solve.
That is why local context matters. For AI to be truly inclusive, it must learn the language of township and rural economies. It must see spaza-shop turnover not as statistical “noise” but as evidence of resilience, entrepreneurship and economic potential. Financial inclusion is not simply about opening accounts; it is about recognising and valuing the diverse ways people already manage money in their communities.
South Africa’s digital infrastructure is gradually enabling this vision. According to FinMark Trust (2022), more than 77% of South Africans now use digital financial services, with 68% accessing them via smartphones. With near-universal 4G coverage and expanding 5G networks, mobile banking has become essential, especially in rural and township areas where physical bank branches remain scarce.
But connectivity is only the starting point. What makes it transformative is intelligence, the ability to turn raw data into personalised, human-centred experiences powered by AI. TymeBank, now with more than 10-million customers, uses AI to tailor financial products and simplify onboarding via kiosks and mobile apps. Similarly, MTN’s MoMo and FNB’s eWallet depend on AI to detect fraud patterns, protecting small traders and informal merchants. These innovations mean someone in a remote village can now store value, send remittances and build a financial identity, sometimes for the very first time.
The economic potential is immense. The World Economic Forum estimates that AI could reduce the cost of providing financial services by up to 50%, unlocking access for millions who were once considered “too expensive” to serve. Capgemini found that three-quarters of banking customers believe AI-powered tools help them make smarter financial decisions. For small businesses, AI-driven platforms can enable real-time inventory tracking, adaptive pricing insights and access to digital credit, helping micro-entrepreneurs transition into the formal economy without the usual red tape.
Africa, often seen as a late adopter of global technologies, is in fact leading a quiet revolution. From Kenya’s M-Pesa to Ghana’s Zeepay and Nigeria’s OPay, AI-enabled ecosystems are redefining financial inclusion. South Africa can learn from these examples by fostering open data collaboration between regulators, telcos, and financial institutions. Shared data sandboxes and ethical AI partnerships could enable more accurate, context-aware risk assessments while protecting privacy and consent.
However, governance will determine whether AI’s promise becomes reality. The question is no longer whether AI will shape financial inclusion `— it already does — but how do we ensure it does so responsibly? Who designs these systems? Who ensures they are fair, transparent and culturally sensitive? Without ethical guardrails, AI may unintentionally widen the digital divide through hidden bias.
Regulators such as the South African Reserve Bank and Financial Sector Conduct Authority must go beyond compliance. They should promote experimentation under ethical frameworks, requiring banks to test AI models for bias, explainability and representativeness, not just profitability. Public-private partnerships could help build inclusive data ecosystems that recognise informal economic activity as legitimate, rather than invisible.
Looking ahead, AI in banking must evolve from automation towards empathic intelligence, the ability to understand and respond to human context. Financial inclusion is not merely a data problem; it is a relationship challenge. For technology to succeed, it must earn trust in communities historically marginalised by formal finance. That means designing digital products that speak local languages, respect cultural practices and strengthen, rather than replace, human connection.
In the end, AI’s greatest promise in banking the unbanked lies not in computation but in compassion. When algorithms are guided by empathy and equity, they become more than tools of efficiency; they become instruments of justice. If we can teach AI to see the value in a spaza shop’s daily sales or a grandmother’s stokvel contribution, we are not merely digitising inclusion; we are dignifying it.
That is the future South Africa and Africa must aim for: an intelligent, fair and deeply human financial system.
- Dr Mahlangu serves on the strategic advisory board of the DaVinci Institute, and Monatisa is a director of the Madisebo Foundation. They write in their personal capacities.
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.