Made in Africa: An African Perspective to the Design, Deployment and Governance of AI

Published: 14 February 2024
Authors: Joshua Baru, Babra Chege, Shikoh Gitau (PhD), Kavengi Kitonga, David Lemayian, Wanjiru Mburu (PhD), Tevin Mwenda, Njeri Ngaruiya (PhD), and William Powers

Summary​

This paper argues that instead of focusing on regulation, which is premature given the current state of its AI ecosystem, Africa should prioritise developing a robust operational AI ecosystem. The authors use the "Four Horsemen" framework to assess the state of AI in Africa and guide development, focusing on:

1. Data Sets and Data Systems: Africa possesses a wealth of untapped data. However, limitations in data sets, like the lack of representation of people of colour, result in biased AI systems and hinder the development of AI products tailored to the African market. Initiatives focused on creating and annotating African data, particularly in areas like Natural Language Processing (NLP), are crucial for creating culturally relevant AI solutions. However, data quality and ethical data collection practices are paramount.

2. Digital Infrastructure: Building robust digital infrastructure is essential. The paper emphasises:

   • Data Centres: Africa needs significant investment in data centres to avoid its data being stored and processed elsewhere. While investments are increasing, the number of data centres remains low compared to other regions. The focus should be on sustainable data centre solutions using renewable energy sources.
   • Connectivity: Despite initiatives like the 2Africa subsea cable system and satellite internet, internet penetration in Africa remains low. Bridging the connectivity gap is crucial for wider AI adoption.
   • Chip Manufacturing: While Africa is absent from global conversations about chip manufacturing, the authors believe it has the potential to become a powerhouse in this domain with the right business and legal frameworks to attract investment. Local initiatives demonstrate this potential, but significant investment is needed.

3. Talent: Africa has a growing pool of skilled professionals, but formal training in AI-related fields is crucial for harnessing their full potential. The paper highlights the need for upskilling initiatives and rethinking labour laws to protect workers' rights. It suggests exploring unconventional training avenues, such as data labelling, to create job opportunities and position Africa as a significant player in the AI market.

4. AI Markets and Use Cases:

   • Awareness: Raising awareness about AI's potential to address local challenges is paramount.
   • Usefulness: AI solutions must be tailored to African realities and address specific challenges like rural development, low literacy levels, and financial inclusion.
   • Commercial Viability: The paper highlights the importance of frugal design innovation, focusing on niche markets, and attracting investment in AI startups to ensure commercial viability and sustainability.

The paper concludes that regulation, while important, is not currently a priority. Instead, Africa should prioritise initiatives that address the identified bottlenecks within the AI ecosystem to become an AI powerhouse. Future articles will expand on each component of the "Four Horsemen" framework, providing a deeper understanding of the current state of AI in Africa and guiding the development of an effective AI regulation readiness framework.

Audio Overview​

Full PDF​