This time last year, Web3 and the Blockchain were the fad and if you were not clued in, you were seen as missing out. Never mind that its use cases were not so defined.
Fast-forward one year and AI is the new kid on the block. At every turn, you're bombarded with tweets from people promising you that you'd be out of business if you don't get on the AI train. Not even escaping to LinkedIn brings respite.
Unlike web3 and its siblings, AI has clear use cases and some of its applications hold significant potential. My colleague, Bolu, recently wrote an article about AI's potential applications. Today, I'm exploring some applications of AI in the fintech space, its drawbacks and challenges.
What is artificial intelligence?
I'd set the stage by attempting a definition of artificial intelligence with the help of its poster child, ChatGPT. According to ChatGPT, artificial intelligence (AI) is the ability of machines to perform tasks that typically require human intelligence, such as recognising speech, making decisions, and solving problems.
AI can do this because it is fed data that help it discover and analyse patterns. As I mentioned earlier, AI can be used in literally any field to identify patterns and make decisions. For example, an AI chatbot can act as a robo advisor, helping individuals decide what asset to invest in.
How can African fintech startups use artificial intelligence?
The first question I had after thinking of possible applications of AI in fintech was whether any startup was already working with artificial intelligence. Africans tend to be on the consuming end of such innovations but fortunately, there are a few startups already using AI to figure out things like credit scoring and fraud detection.
Victor Irechukwu, Head of Engineering at OnePipe confirmed this stating that OnePipe already uses artificial intelligence for fraud detection and prevention and biometric verification for customer KYC implementation. It also uses natural language processing for conversational transactions.
So how can artificial intelligence be used in fintech? Irechukwu identifies five major areas fintechs can explore — biometric verification, compliance and regulatory reporting, customer acquisition and retention, credit scoring and risk assessment, and finally, fraud detection and prevention.
Babatunde Akin-Moses, CEO of fintech startup, Sycamore adds two areas — personalised marketing and securities trading.
The five areas mentioned by Irechukwu are major headaches for any fintech startup. Take credit scoring and risk assessment, for example. Nigeria has a low credit penetration and despite the best efforts of fintech startups, many Nigerians are still unable to access credit.
According to Irechukwu, AI can analyse customers' data to evaluate creditworthiness and assess risk, which can help lenders make more accurate decisions and reduce the risk of default.
On the other hand, fraud detection and prevention could do with AI's help. In March 2023, Techpoint Africa reported that Flutterwave had lost huge sums to a hack. Reports have since emerged showing that fintech startups are working on a model for fraud prevention.
A recent report by Kaspersky also claims that 37% of Nigerians have lost money while using online banking channels. AI could help identify and prevent these fraudulent activities.
What challenges would fintech startups face when using artificial intelligence?
It's easy to get the sense that using AI could solve all the problems that fintech startups currently face, but that's not the case. There are a few challenges that come with the territory, with access to data, a challenge mentioned by all the experts I spoke with.
Yvonne-Faith Elaigwu, Head of Operations at OnePipe explains that access to data could be a major challenge for fintech startups and shared an experience with a fintech startup she worked with in the past.
"You need to feed the model with sufficient data so it can identify patterns and make decisions. I recall working with a company some years ago that was trying to use AI to help lenders make lending decisions.
"The biggest hindrance was access to data because a lot of people did not have sufficient financial "digital footprint" from which information could be obtained. Hopefully, the proliferation of fintechs working to bank the unbanked and underbanked across Africa, will somehow solve this issue of access to data."
We have a running joke in the Techpoint Africa editorial team about the challenges of finding data but more importantly, finding accurate data. Irechukwu buttresses Elaigwu's point.
"AI models require large volumes of high-quality data to be effective. In the fintech space, there may be challenges in accessing quality and managing this data bearing in mind that AI models are only as good as the data they are trained on," he says.
He adds that fintech startups would also have to combat regulatory considerations when deciding how to use AI in their operations, while they must be able to explain its workings to the relevant stakeholders.
So far, African startups have not taken the lead in AI but given these are still early days, there's a chance they could play a huge part going forward. "While we might not be at the forefront of developing the technology itself, we can certainly find creative use cases for it. As with most technologies, it's those who find the most practical and useful applications that win the market. Not necessarily those who develop the underlying technologies. I'd say African fintechs can certainly take part in the conversation in a very meaningful way," Akin-Moses shares.