Fake AI 

Edited by Frederike Kaltheuner

Meatspace Press (2021)

Book release: 14/12/2021

This book is an intervention - 

Chapter 14

When fintech meets 60 million unbanked citizens

By Favour Borokini and Ridwan Oloyede

In Nigeria, AI is being touted by some as a one-size-fits-all solution to the country’s inefficiencies and woes.1

An unfortunate combination of sclerotic (and occasionally regressive) domestic development with Western-influenced tech solutionism has resulted in a burgeoning fascination with the technology. Nowhere, perhaps, is this more readily evident than in the fintech industry.

Fintech, a portmanteau word that describes the integration of financial services with technology, is often deployed by companies offering financial services to penetrate new markets. Following a series of high profile funding successes, especially that of the “country’s latest unicorn”,2 the term has become a buzzword in Nigeria.

According to a 2020 McKinsey report,3 Nigeria is home to over 200 standalone fintech companies. This figure does not include the dizzying array of fintech offerings by Nigerian brick-and-mortar banks. Yet, Nigeria is one of seven countries that contribute to nearly half of the world’s unbanked population, totalling about 60% of her adult population.4 In part, this is due to the sheer number of people living in severe poverty (83 million compared to India’s 73 million).5 This is exacerbated by a conservative and seemingly erratic economic policy and a lack of access to financial services, particularly in rural areas.6 With the median age of Nigerians being around 18 years, home-grown fintechs are stepping up to the challenge of providing financial inclusion services to its youthful population.7 By using targeted advertisements, appealing offerings, and other digital marketing strategies, fintechs may appear to have a better chance of reaching these young people than conventional banks.

Certain Nigerian fintechs, which offer their services to individuals or corporate clients, claim to leverage “machine learning algorithms” and “AI-powered facial recognition” to assess credit risk, prevent fraud, generate personality profile reports, and for identity verification.8 Several fintechs use these products. A closer look at the environments in which these solutions are deployed raises serious questions about whether they can deliver on their promises to improve financial inclusion. Instead, the real risk seems to be that these solutions could exclude minorities and privileged profit at the expense of individuals’ rights and liberties. To identify and profile the ideal customer, these companies typically access the personal data of private individuals from the government, through third-party service providers, or harvest it directly by sifting through personal information on mobile devices, such as text messages, fine and coarse location, media contents, contact lists, social media, and use of trackers and other permissions. One company, which promises to empower Africans “by driving social and financial inclusion” currently pays an undisclosed amount to Nigerian government agencies to gain access to millions of Nigerians’ national identity data.The company’s publicly stated goal is to promote trust through digital identity and verification services. Another company claims to help “banks distinguish between a photograph in an ID and a selfie”, to which end it has “created an identity management system that harnesses the powers of facial recognition technology.” This system connects to government databases and non-government databases, such as that of the Nigeria Inter-Bank Settlement System. Both companies promise that AI-powered facial recognition systems will improve integrity and better identify fraud by “knowing all details about the client in near real-time.”

The kind of access required by these systems is incredibly invasive. Ultimately amounting to surveillance of the activities of private citizens, it is potentially in breach of rights enshrined in the constitution and other laws, including freedom from discrimination, privacy, data protection and dignity. As other authors in this book have argued, these systems, which are not known for their accuracy or fairness, could be unfairly prejudiced against persons based on their gender, socio-economic background or other discriminatory factors. More troubling is the prospect of fintechs ending up as gatekeepers, shutting out the excluded groups they claim to include. If, for example, an individual is unable to purchase a smartphone (or electricity or internet to make use of it) and one of these systems, therefore, cannot automatically extract data, it will rank them as undesirable, further excluding them from access to credit and financial products.

According to a report published by Tech Hive Advisory about the pervasive practices of digital lenders in Nigeria,10 seven of the 22 mobile applications analysed publicly disclose that they use AI to determine borrowers’ creditworthiness. Only one mentioned the existence of profiling in its privacy notice, as is required by law.11 Besides this disregard for the law, the report also notes an alarmingly common lack of algorithmic transparency, lack of explainability, lack of accountability, and an absence of information on purpose limitation of the data used.

It appears that the solutionism being sold by fintechs and other players like digital identity providers, namely the belief that AI can single-handedly fix structural deficiencies, has been bought hook, line and sinker by those who ought to be the last to do so — the government. Neglecting or outsourcing civic obligations (such as digital identity registration, and in particular, the development of robust financial inclusion policies) to profit-driven private enterprises running machine learning algorithms without sufficient safeguards will undoubtedly worsen already-existing inequalities by systematically breaking down civil rights and freedoms. One such example is the licensing of private entities to access national identity biometric data held by the National Identity Management Commission, some of which claim to use artificial intelligence and facial recognition.12 The details of the agreement are not available publicly. No record of a data protection impact assessment being conducted has either been publicly stated or published. These and other machine learning algorithms that make predictions based on historical events and data cannot be at the forefront of providing the forward-looking information we need for our future.

While AI has its uses in industry—fintechs included—it cannot replace a comprehensive financial inclusion and development approach. Certainly not when spurred on by a lack of transparency and accountability. The performance, validation and deployment of AI must be ethical and meet existing legal requirements. Fairness and transparency must determine limits to how data is used and the algorithms that are deployed. Security, privacy, data protection, and accountability about how data is used and by whom is critical. Most importantly, it is essential to understand users, their needs, and the context in which these technologies are being used.

Favour Borokini is a tech policy researcher interested in (emerging) technology-facilitated violence against women and the development and deployment of AI in Africa.

Ridwan Oloyede is a Co-Founder at Tech Hive Advisory, where he focuses on global data protection and privacy laws, digital health, and digital ethics, among other issues.


1. Editorial Board. (2021, March 8). Nigeria Needs Artificial Intelligence to Combat Insecurity, Says Expert. The Guardian Nigeria. https://guardian.ng/news/nigeria-needs-artificial-intelligence-to-combat-insecurity-says-expert

2. Jackson, T. (2021, March) Nigerian Payments Startup Flutterwave Achieves ‘Unicorn’ Status after $170m Funding Round. Disrupt Africa. https://disrupt-africa.com/2021/03/10/nigerian-payments-startup-flutterwave-achieves-unicorn-status-after-170m-funding-round/

3. Kola-Oyeneyin, E., Kuyoro, M., & Olanrewaju, T. (2000, September 23). Harnessing Nigeria’s fintech potential. McKinsey & Company. https://www.mckinsey.com/featured-insights/middle-east-and-africa/harnessing-nigerias-fintech-potential

4. Ventura, L. (2021, February 17). Global Finance Magazine - World’s Most Unbanked Countries 2021. Global Finance Magazine. https://www.gfmag.com/global-data/economic-data/worlds-most-unbanked-countries

5. Forty percent of Nigerians live below the poverty line: Report. (2020, May 4). AlJazeera.https://www.aljazeera.com/economy/2020/5/4/forty-percent-of-nigerians-live-below-the-poverty-line-report

6. Osakwe, S. (2021, April 6). How Is Nigeria’s National Financial Inclusion Strategy Going? Center for Financial Inclusion. https://www.centerforfinancialinclusion.org/how-is-nigerias-national-financial-inclusion-strategy-going

7. Roland Berger Strategy Consultants. (2012, January). National Financial Inclusion Strategy. Central Bank of Nigeria. https://www.cbn.gov.ng/Out/2012/publications/reports/dfd/CBN-Summary%20Report%20of-Financial%20Inclusion%20in%20Nigeria-final.pdf

8. Onukwue, A. O. (2020, November 19). The BackEnd: Meet the “Palantir of Africa.” Techcabal.https://techcabal.com/2020/11/19/the-backend-analytics-intelligence-palantir-africa/

Hersey, F. (2020, August 5). Verify my life: could a Nigerian problem lead to a global trust solution? (Or fuel a two-tier society?). Biometric Update. https://www.biometricupdate.com/202008/verify-my-life-could-a-nigerian-problem-lead-to-a-global-trust-solution

10. Tech Hive Advisory (2021, February) Digital lending: Inside the pervasive practice of LendTechs in Nigeria. LinkedIn. https://www.linkedin.com/posts/tech-hive-advisory_digital-lending-inside-the-practice-of-lendtechs-activity-6768431134297620480-E5zx 

11. Article 3.1(7)(L) of the Nigeria Data Protection Regulation (NDPR).

12. National Identity Management Commission (2020, 15 December) Public Notice: Approved Data Capturing Agents (Digital Identity Ecosystem) [Press release]. https://nimc.gov.ng/public-notice-approved-data-capturing-agents-digital-identity-ecosystem

Next: Chapter 15
Algorithmic registers and their limitations as a governance practice

by Fieke Jansen and Corinne Cath

Instagram        Twitter