Before Your African Bank Deploys Artificial Intelligence (AI)

Artificial intelligence (AI) is hot, but sub-Saharan African banks should get over the hype. There is need to do a thorough synthesis analysis before spending money on this new technology. Personally, at this phase of our banking development, deploying AI across different business units will have only marginal impacts when evaluated with the deployment costs. Practically, it makes no sense, for a bank to spend so much on AI because we are not ready. I understand the excitement that comes with leapfrogging challenges with technology; AI will not offer such benefits in our capacities to fix key banking business frictions, at scale, at this moment.

Sure, many consulting and technology firms are visiting banks across Africa running demos on how AI could magically grow revenues. That will happen but not anytime soon. Besides our lack of data and depth in understanding market patterns, AI (especially those engineered outside Africa) will struggle to add meaningful value.

Anyone that tells you that he has figured out how AI, especially ones created outside Nigeria, will trade stocks in Nigeria and return huge returns, you should tell the person to start a malaria treatment [AI can help in research, but not in autonomous trading]. No one has that capability because no one has the data to test such capabilities. The Nigerian Stock Exchange and the Securities & Exchange Commissions may not even have (complete) trading data that is more than ten years old in usable electronic formats. So, all these models are largely new and cannot be relied upon. Places like U.S. and some EU regions have data they have accumulated over decades, making it possible to build and test models with higher level of accuracy. Besides, their economic structures have largely matured – they are heterogeneous economies while most African economies are homogeneous economies which make us more susceptible to trade shocks, arising from price-gyration of commodities. With minimal exceptions, Germany and UK markets are more closely related than Nigeria and South Africa, or Kenya and Gabon, with our minerals and hydrocarbons playing dominant roles.

Some AI Applications in Banking

There are many ways AI can help in our banking sector. But in some of these areas, one does not need AI to deploy contemporary IT solutions in the noted business frictions. In other words, even before AI, we should have used current IT solutions to address them. The following are areas AI has promise:

Anti-money Laundering and Fraud Detection: The use of pattern recognition technology can improve anti-money laundering and fraud detection activities. Even though we like to throw AI into this, any bank not doing this now is not really using IT. Sure, it has to get better and AI can help. I do not see any risk in deploying AI in this  area. This is not a concern.

Chat bots: A bank can use chatbots but I am not sure if customers are ready to engage bots with their financial lives on digital ecosystems. The trust is not there yet. Any bank doing this may launch it but most bank customers will not get close to using it. Yes, while the technology can be built, the customers may not use it.

Algorithmic Banking:  Data and analytics can help business leaders see business patterns, understand their firms, and drive allocation of capital. Any bank should be doing those things. But using autonomous algorithms as they do in NYSE to trade stocks and other trading services will be careless in Africa (except South Africa where they have data) today. This is where I have the biggest concern on the application of AI in the African banking sector.

Recommendation Technologies: Any bank with data today should use IT (even before the transition to AI) to drive its lending, mortgage business and more. AI can indeed help but no one should wait for AI  before it can use common IT tools to understand its business. Every bank has data of its customers, and using that data to make future decisions can be done with simple rudimentary solutions in the domain of IT. I expect every bank to be doing this at the moment.

All Together

Yet, while I do not see any promise in using AI to drive investment and market-moving decision including lending at scale, AI can help in improving customer insights. If you are a broker, AI can assist you to support customers by analyzing their portfolios, helping them to balance their portfolios through optimized asset allocation strategies. Yes, you should not be focusing on how to use AI to beat the markets because you cannot do such, as you have no data to test such models. Just as we cannot do (autonomous) AI-driven lending without massive datasets of customer credit histories, we will struggle in building trading models without trading data.

AI has a promise to connect business elements (credit: Udemy)

It will be irresponsible for any bank or financial institution to link products to its core general ledger expecting AI to execute financial transactions in Africa autonomously. The problem is not the capacity to make such algorithms, but the data to examine that they make sense. To build an investing system to allow AI to trade in any African exchange is reckless at the moment. I saw a promo by a company promising that its AI can beat Nigerian Stock Exchange. I was like: how did they test the model? Who provided the data? It is possible they have the data, and if they do, that may be the most important innovation because that data will certainly give them a great competitive advantage. But I doubt it – it is likely a product made for NASDAQ and NYSE massaged for the Nigerian Stock Exchange with some fudge factors.

Finally, I am not saying that AI is not useful: sure, it will add value in our banking. My point is that we are not there yet especially when it involves beating markets with models. We can continue to use information technology to improve our banking operations but the transition to AI must be carefully executed. There are many productivity gains which IT offers even without the elevation to AI. We have not totally exploited those gains. The time for AI will come, as we build data, and new opportunities will emerge.

---

Register for Tekedia Mini-MBA (Feb 10 - May 3, 2025), and join Prof Ndubuisi Ekekwe and our global faculty; click here.