A roadmap for artificial intelligence for development in Africa

Based on the workshop ” Toward a Network of Excellence in Artificial Intelligence for Development (AI4D) in sub-Saharan Africa” – April 3-5, 2019 Nairobi, Kenya.

 

Artificial intelligence is poised to enhance productivity, innovation and help countries across sub-Saharan Africa to achieve the Sustainable Development Goals – helping to improve outcomes from health care to agriculture to education. Yet as with any technology, the transformative potential benefits come with challenges that need to be managed and moderated. Though countries across sub-Saharan Africa are on this threshold of harnessing AI technologies to support genuine human development and bolster economic and political progress, there are likely to be ramifications on already precarious livelihoods, labour markets, and fragile governing institutions. Addressing these opportunities and challenges are key to the roadmap for AI for development.  

The AI Readiness Index 2018 (forthcoming)
The AI Readiness Index 2018 (forthcoming)

Infrastructural challenges, data gaps, highly skilled labour needs, and poor regulatory environments are still inhibiting people’s ability to harness AI across Africa. A global review of secondary data available on AI suggests that the continent needs to support better infrastructure and fill data gaps, and support education and capacity building to ensure there are enough highly skilled researchers able to implement AI solutions. Moreover, a critical dimension is the legal, ethical and human rights-based frameworks needed for countries to optimize the best uses of AI – frameworks that will protect citizens’ data and build trust and legitimacy for AI and the multitude of applications. These are still largely absent.

In the next five years, what would happen if we had more than 30 African countries developing their own AI policies and strategies? Imagine if there are more than 400 PhDs in AI and machine learning from across the continent? And what if universities, the private sector, and other public interest institutions invest a billion dollars in collaborations on AI to support the African sustainable development agenda?

These proposals are now emerging from the African AI community after the recent workshop in Nairobi on Artificial Intelligence for Development in Sub-Saharan Africa. The meeting gathered sixty African and international experts together to demonstrate and discuss how people across the continent are developing AI at a rapid pace. There is a vibrant AI ecosystem emerging across the continent with initiatives like Deep Learning Indaba, the African Institute for Mathematical Science (AIMS), Data Science Africa, and Women+ in Machine Learning and Data Science (WIMLDS) — all of which aim to strengthen machine learning and artificial intelligence in Africa.

To support the growing momentum in Africa’s emerging AI ecosystem, this emerging ‘network of excellence’ of machine learning and AI practitioners and researchers is building a collaborative roadmap for AI for Development in Africa. The three-day workshop focused in on three critical areas of 1) policy and regulations, 2) skills and capacity building and 3) the application of AI in Africa. The following slides and the analysis below synthesize the participatory discussions during the workshop on these three critical areas.  

Artificial Intelligence Network of Excellence in Sub-Saharan Africa – Nairobi Workshop 2019

Policy and regulations

“Thirty African countries should develop AI specific policies and strategies over the next five years”

Currently, out of the 46 sub-Saharan African states, only Kenya has an AI taskforce that is working towards a national strategy. There is a clear need to build policy capacity as well to help design AI regulatory frameworks fit for the African context. An area of convergence in this discussion was the need to channel investments toward prioritized solutions for development, but also the need to reflect and better understand what “development” means. The pan-African development blueprint – the African Union’s Agenda 2063 – could offer such a shared vision. Other ideas emerged during the workshop, including collaboration within the network to inform and grow AI-specific policies to at least 30 African countries over the next five years, and increasing the adoption of the AU’s Convention on Cyber Security and Personal Data Protection. There is also critical recognition that the copy-and-paste of made-in-the-North AI policy frameworks will not serve the needs of Africans. No one country has all the solutions. So while policy options and best practices from other parts of the worlds can provide great sources of inspiration and guidance, the priorities and directions must be determined locally in a way that facilitates innovation, curbs any harms and upholds human rights.

Skills and infrastructure

“Create a pipeline of 400 African PhDs in AI, data science, and other interdisciplinary fields.”

Building an inclusive AI future in Africa means we must build AI strength through community. Skills and capacity received a lot of attention during the three-day workshop. One of the participants at the workshop shared that there is a need to “better formalize AI training and the recruitment of AI talent in Africa” and to create a pipeline of 400 African PhDs in AI, data science, and other interdisciplinary fields over the next five years. This means having the appropriate infrastructure to build this capacity.

Furthermore, there is the need to enhance the strength and agility of educational systems to meet new digital opportunities, and support investments in learning outcomes in the areas of machine learning, artificial intelligence, and data sciences in Africa. As one participant said, “Let’s think beyond PhDs. We should also target and invest in youth at the primary school level, offer mentorship programs for emerging leaders, including for women and girls to build the next generation of AI practitioners, and create opportunities to bring more of the public into the AI conversation.” Moving forward, inclusion must remain a core principle in guiding the design of responsive and equitable AI skills and capacity building roadmap in Africa.

The hope is that it will be possible to establish an AI Centre of Excellence in every African country by 2030 that will help incubate ideas and foster AI communities of practice in an interdisciplinary and inclusive matter.

Applications

“A collective investment of USD 1 billion dollars in collaborative innovation and research prioritizing solution areas for sustainable development in Africa.”

The application of AI must be connected to people and their needs on the ground, and the potential benefits of AI translated into real impact for people. The ground-up approach of this collaborative workshop was an excellent way to start drawing fresh ideas and practical use cases on how AI could help increase access to healthcare and education for the most vulnerable in rural areas, and improve the movement of people, goods, and ideas especially within and between rapidly growing megacities across Africa.

A key idea discussed among participants was how to mobilize collaboration among a network of African companies, universities, research centres, and public institutions to collaborate on advancing the AI research for development agenda. Participants suggested a collective investment of USD 1 billion dollars in collaborative innovation and research prioritizing solution areas for sustainable development in Africa. These partnerships could mean when Africans design and deploy AI applications, societal goals and human rights commitments like decent work conditions and gender equality are integrated into projects from the beginning.

Next steps

It is very clear from the diverse range of voices at the workshop that we cannot forget that within Africa’s emerging and vibrant AI ecosystem, the inclusion and diversity of voices from traditionally marginalized communities must be prioritized. And there is a need to expand the AI Network of Excellence conversation to Francophone and Portuguese-speaking countries on the continent as well. It is also critical to understand that contexts differ in Kenya, Nigeria, and South Africa and even more so in places where AI is still unknown, so strategies and engagement must be tailored to local situations.

The enriching and collaborative dialogue over the three days of the AI4D workshop has built a solid foundation for the research and innovation agenda in Africa. Progress on this agenda will depend on the collective action of many actors to hone and support this vision, and we will continue to engage with partners and stakeholders in shaping an inclusive AI ecosystem in Africa.  

See also:

 

We would like to thank all the participants who contributed their voices and ideas during this event. This three-day dialogue was organized by the International Development Research Centre, Swedish International Development Agency, Knowledge for All Foundation and the Strathmore University in Nairobi, Kenya. Follow us on Twitter at @AI4Dev