The buzz surrounding artificial intelligence (AI) in East Africa’s financial sector is undeniable. Yet, when I speak with regional banking leaders, there is a subtext of caution. While the excitement for efficiency is real, there is a growing realisation that for AI to work in our unique context, the ground beneath it must be solid, ethical and deeply rooted in the local reality.
The most urgent question facing East African banking today is not how quickly we can automate, but how responsibly we can embed these technologies into our institutions.
In a region where banking is built on hard-won trust and personal relationships, AI must be an augmentation tool rather than a substitution strategy. This is not just an ethical stance; it is a business imperative. Our markets are high-risk and high-trust, and AI cannot succeed unless our clients understand it.
In East Africa, responsible AI begins long before a single line of code is written. We must ask ourselves what problems we are truly solving. Is the goal to drive financial inclusion for millions of unbanked Kenyans, Tanzanians and Ugandans, or is it merely to save costs?
We must be clear about the data driving these models. In a region where formal credit histories are often sparse, the data we use must be fair and representative. Without this clarity, even the most advanced systems become liabilities.
We must also stop tiptoeing around the challenge of bias. Bias is not just a technical glitch; it reflects our history and our social systems. In East Africa, where ethnic, gender and geographic differences can affect access to money, AI might make these biases worse.
Statistics from the World Bank and other regional central banks often highlight gaps in credit access for women-led Small and medium enterprises (SMEs) and rural farmers.
If an algorithm is trained on historical lending data that favours urban, male borrowers, it will naturally penalise others. We must take an engineering-first approach to this, using diverse datasets and explainable AI tools that uncover the "why" behind a loan rejection or a credit limit.
The current landscape also reveals a governance vacuum. Many regional institutions are encouraging their teams to "lean into AI" without providing the necessary guardrails. This creates risks ranging from the misuse of sensitive customer data to the over-automation of judgment-driven decisions. We do not necessarily need more AI; we need better governance.
Good governance requires clear playbooks on where AI can operate and, more importantly, where it must stop. For high-stakes decisions, human oversight must be a nonnegotiable part of the workflow.
We also have a unique opportunity to bridge the gap between our legacy banks and our thriving fintech ecosystem. Legacy banks in East Africa bring decades of regulatory experience and deep customer trust, but they struggle with integration.
Fintechs bring the agility of "Silicon Savannah," yet they eventually face the challenge of scaling governance. Both must borrow strengths from each other to create an AI ecosystem that is transparent and accountable.
Thriving financial institutions will be those that treat AI not as a shortcut, but as a long-term commitment to responsible innovation. The path to a prosperous digital future for East Africa begins with a simple, foundational belief: AI must remain human at its core.
Furthermore, we must change how we communicate AI to our own people. If a bank teller in Nairobi or a credit officer in Entebbe feels that AI is a threat to their livelihood, they will resist it.
However, if they see AI as a tool that handles low-value, repetitive tasks, leaving them free to focus on complex relationship management, they will embrace it. Transparency is the only way to build this internal trust.
The writer is the Managing Director of NTT DATA, East Africa
