Saturday 2nd May 2020, a day that may go down in history as the beginning of Africa wide collaboration of AI communities. This was the day 14 AI communities, representing 7,400 members came together to brainstorm and strategize what an Africa wide AI community should look like, and what it should focus on achieving. If you are already sold on this idea, scroll to the end for next steps. Otherwise, let’s tell you why we met.
Artificial Intelligence has been a difficult discussion in the developing world due to the perception that we have more pressing challenges to attend to and something like AI is too advanced for us to even consider. And those who recognize AI’s value in taking us a step closer to solving our challenges face significant hurdles.
This notwithstanding, several AI communities sprung up in the 2018 period, where enthusiasts and practitioners would meet to exchange best practices, pool resources to achieve greater goals and create a path for a new generation of African AI Talent. Each community has carried the torch to light the way of AI’s potential in their home nation, with each forming an expertise in a particular field such as content creation, project management, product creation, and hackathons. Now it is time to bring together the flames and create a beacon for Africans that will feed into the growing flame of Africa’s development.
The first step to any collaboration is identifying the challenges we face as a collective and what outcomes we want to achieve. Hence we ran a brainstorming call with top communities to hear directly from them.
The top 5 challenges that stood out were
Little collaboration in the Ecosystem
Being able to work with the practitioners and stakeholders brings a powerful understanding of local context, and when this understanding is multiplied over the nations in the continent we create a deep understanding of what opportunities and challenges exist and who is capable of working to solve them.
“You’re in a country or anywhere, but then you don’t have the full technical know-how to do it but you start, and then you get stuck somewhere. …. so if there is a collaboration with other people with the skill and others with the domain expertise, you can seek help from them.”~ Martha Teye, Zlitch Technologies
Being able to bring together those who have the expertise, the data, the technology, and the challenges, has shown to yield amazing results. Now we must foster this culture in the AI Communities by bringing together the African industries, academia, civil society, and policymakers.
A lack of job and internship opportunities
There is a growing desire from learns to apply their knowledge to real problems, so they grow more confident to participate in global competitive markets.
The lack of the local internship, job opportunities and even mentor-ship make it challenging to prepare students for work environments beyond those in university projects and start-ups.
“Most of our students are doing AI research. This requires supervisors who actually understand what they’re doing because you might have a title and you start studying about it and you want to work on it, but lack a supervisor that would mentor and direct you on that topic. So you get a supervisor that is not well versed with AI. And so you don’t get good feedback that will help you improve your work.” ~ Maria from The Nelson Mandela African Institution of Science and Technology
A lack of access to computing resources
It is near impossible to build competitive AI solutions without access to powerful compute infrastructure, like GPUs.
The rise of cloud computing is reducing this hurdle day by day, as said by Maria, “Tools like the Google collaborator, why don’t we use the Internet, and run it somewhere else, that’s another way of solving the GPU issue. But then the internet is very expensive. So that was not really a solution to most of the students.”
The ecosystem could also benefit immensely from guides on how to access and build local compute clusters, as this route is generally more suitable for learning and experimenting, before deploying ready solutions on the expensive cloud option.
A lack of localized accessible data
You’ve probably heard the saying, “Data is the New Oil”, and this is especially true for any AI or ML product. Sure we can use internationally sourced data to create cool projects and prototypes, but at the end of the day if we lack contextually accurate data for the people we aim to help, then our products will never truly be feasible.
What we do see is that there exists data with similar context from neighboring countries, it takes the use of connections to find what data is there and how we can use it, but exist it does.
If we have a centralized repository of African focused data, not only can we speed up the time to deployment, we can show the value to students on how this technology can be used in our own homes and nations.
Lack of Locally available, certified AI courses
When entire employment systems have been built to prioritize the papers that ‘credit’ us to have the skills we claim, it creates a harsh environment for the self-learners who gain their skills through the likes of YouTube, community meetups, and online coding challenges.
The options we have, wait for the majority of industries to adopt a culture that is accepting of self-learners who show their skill through their repositories or find ways to offer cheap courses that can be certified by bodies that carry authority in the industry.
Another challenge highlighted by Rose Delilah from Nairobi WIMLDS “…..You find we have individuals of different levels. We have beginners, intermediate, and advanced. And the AI industry is broad you have people who are dealing with natural language processing and you find we have few people who can really deep dive educate”
Honorable mentions:
- Lack of an African AI Vision: A fascinating point to ponder on came up if we could only choose a handful of sectors to apply AI in Africa, which should we choose?Mohammed Khalil, the AI professor from MoroccoA made a call for a clear vision and focus on the development and application of AI in Africa, derived from the priority application in each of the countries which have their own development agenda.
- Lack of Funding: Funding is a challenge that is often there, and rightly so when we hope to have people to teach skills, and require a place and the resources to enable the teaching. While some are fortunate to have access to funding opportunities, it is not as easily accessible to all, especially when it is most required in the formative phase where a community is getting on its feet and recruiting the skill and resources to grow. As we moved closer to the end of the session with a clearer understanding of the challenges we need to work together to address, we began to consider how we could dive deeper into the challenges and learn the nuances as well as our individual strengths and actions we can take to contribute to a solution.
Thus the validation for the first inter-community virtual conference was in place, a platform that would offer a variety of activities designed to help all participating communities grow, from the exchange of ideas and expertise. We have all co-designed this event to be a mix of expert sessions and discussion break-out rooms, where participants come prepared to contribute in areas they have excelled, and learn in others where they need guidance.
We welcome you to complete this sign up form for the first virtual conference on May 30, if you lead an AI community in Africa and believe that we can grow together. You will receive further guidance via email, and in the interim can sign up to the Alliance4ai newsletter to stay updated on everything AI in Africa.
If you have general questions or are a media, technology or banking group interested in sponsoring these activities, send a note to admin@alliance4ai.org
For now, continue to stay healthy, be well!
Conveners
Martha Teye, Zlitch Technologies, Ghana
Essa Mohamedali, TanzanAI Lab, Tanzania
Alex Tsado, Alliance4ai, Nigeria