The AI triad and diffusion pathways provide a lens for understanding the internal dynamics of the AI ecosystem in East Africa. Infrastructure and finance mechanisms are the external factors that enable or hinder the development of the ecosystem. They underscore the importance of physical resources (electricity, data centers) and human capital (researchers, engineers) in driving AI innovation.
Internet connectivity and data generation are inextricably linked, with internet penetration driving the creation of vast amounts of natural data. In 2024, Nigeria exemplified this connection, boasting the largest number of internet users in Africa. Mobile devices have become the dominant force in global internet usage, accounting for nearly 60% of web page views worldwide in January 2024. This trend is even more pronounced in Africa, where over 73% of web pages are viewed on mobile devices. In contrast, North America still relies heavily on desktop connections, with mobile accounting for just 45.5% of web traffic. The sheer scale of internet users in certain countries underscores the importance of connectivity. In January 2023, China led the world with a staggering 1.05 billion internet users, dwarfing the United States' 311 million. Notably, all BRIC countries had over 100 million internet users, illustrating the correlation between population size, internet penetration, and data generation. Despite progress, disparities in internet access and usage persist. Sub-Saharan Africa grapples with the largest coverage and usage gaps, highlighting the uneven distribution of digital opportunities. The maturity of a country's digital economy directly influences its capacity to generate and utilize data, with Kenya (37%) and Rwanda (30%) making strides while Denmark leads globally at 84%. Internet connectivity is a fundamental driver of data creation and a key indicator of a nation's digital progress. When it comes to Satellites which are critical infrastructure for data generation East Africa has a total of 7 Satellites: Kenya has 1, Rwanda 6 Satellites are crucial infrastructure for data generation, and East Africa is making strides in this arena with a total of 9 satellites in orbit. Kenya boasts 1 satellite, while Rwanda leads the region with an impressive 6 satellites. Ethiopia has launched several satellites into orbit. As of July 2024, they have launched two Earth observation satellites.
Language is the bedrock of Natural Language Processing (NLP) development, as it provides the raw data and linguistic structures upon which NLP models are built and trained (LLMs) like GPT-3 rely heavily on massive text datasets, and the majority of these datasets are in English, leading to an inherent bias towards English language understanding and generation. African languages, with their rich diversity and nuanced expressions, are significantly underrepresented in these training datasets compared to languages like Hindi, which benefits from a larger online presence and more digitized content. This disparity in representation hinders the development of accurate and culturally relevant NLP models for African languages. African languages (Swahili, Twi ,Hausa, Venda, Afrikaans, Bambaara & Malagasy) account for 0.02% with Swahili accounting for only 0.00135% against English at 53%, Hindi 0.1%
East Africa is experiencing significant growth in server investment, outpacing the global average by 13%. Countries like Kenya, Rwanda, and Ethiopia are attracting substantial local and foreign investment in data centers, contributing to a total of 19 data centers in the region. Kenya leads this number with a total of 13 data centers.
Despite these positive indicators, East Africa's compute ecosystem faces challenges due to a limited number of cloud and colocation data centers. The region's cloud computing penetration for the Sub-Saharan Africa rate stands at a mere 15%, compared with 71% in Europe predominantly served by foreign providers with data centers located outside the region.
The state of algorithmic development in East Africa can be gauged by examining three key indicators, I borrow this framework from this brilliant paper: knowledge flows, patents, and academic research.
Knowledge flows: Analyzing the movement of algorithmic expertise and information within and into East Africa. This can be assessed by tracking factors such as: International collaborations: The extent to which East African researchers and institutions collaborate with global leaders in algorithm development. Technology transfer: The adoption and adaptation of algorithms developed elsewhere to address local challenges.
Skill migration: The influx or outflow of talent in the field of algorithms, indicating the region's ability to attract and retain expertise.
Patents: The number and quality of patents related to algorithms filed in East Africa serve as a proxy for innovation and commercialization potential. Examining patent data can reveal: Areas of specialization: The specific domains where East African researchers and companies are focusing their algorithmic efforts. Novelty and impact: The degree to which East African algorithms are pushing the boundaries of existing knowledge and solving real-world problems
Academic research: The quantity and quality of academic research on algorithms conducted in East Africa reflect the region's intellectual capacity and research infrastructure. Evaluating research output can highlight Research focus: The specific areas of algorithm research that are prioritized by East African universities and institutions. Publication impact: The influence of East African research on the global discourse on algorithms, as measured by citations and recognition in international journals.Talent pipeline: The development of a skilled workforce capable of advancing algorithm research and development in the region.
This is yet to be empirically explored, but we hope to be able to do this soon.
According to this assessment of digital governance in 44 sub-Saharan African countries using the G5 benchmark which has four pillars:Pillar I: National Collaborative Governance: This measures how well different stakeholders (like government agencies and policymakers) work together on digital issues. Pillar II: Policy Design Principles: This looks at how digital policies are developed and how transparent the process is. Pillar III: Digital Development Toolbox: This assesses the strategies in place to grow the digital economy and ensure these strategies align with the Sustainable Development Goals (SDGs).Pillar IV: Digital Economy Policy Agenda: This measures the frameworks in place to support digital innovation, digital transformation, and fair taxation policies.
Overall, the average score for the G5 benchmark across the 44 countries is 39.96 out of 100. With East African countries ranking as follows:
Rwanda leads the pack with an overall score of 67.90, particularly excelling in national collaborative governance and demonstrating strength in digital development tools. Kenya follows closely with a score of 60.80, showcasing notable progress in policy design principles and leading the region in its digital economy policy agenda. Uganda scores 54.63, indicating potential in collaborative governance but lagging in the development of digital tools. Ethiopia and Tanzania face challenges in their digital economies, with scores of 47.84 and 45.68 respectively.