Earlier this year, I was on a panel at the Transform Africa summit in Kigali titled “Empowering Africa’s digital disruptors.” I spoke about the data economy and how we can build data-driven economies on the African continent. This article comprises my response to some of the questions I was asked and a few additional ideas.

Q. At the heart of accelerating a single digital market is building data-driven economies. Can you talk about how your applied machine learning research supports a data-driven economy?

A. A data-driven economy is one in which data is a core asset of the economy. Data in its raw form is not structured, it is too messy to make sense out of it. You need tools and processes to extract value from it. You need to refine it for it to be useful. And that’s where applied machine learning comes in.

For example, I’ve been working on natural language processing for codemixed contexts. The motivation behind NLP for codemixed contexts stems from the fact that in informal or semi-formal environments, the use of language across the continent is often mixed. In Kenya, for instance, the use English language is heavily influenced by and often mixed with Swahili in semi-formal or informal settings often referred to as Sheng.

You see, in a data economy, those who build the data refineries extract the value and make all the money. As such, in the case of the NLP research, the implications are that we can build and deploy tools like conversation agents, speech recognition that actually works with the way we write or speak as Africans.

As the data economy grows in importance, applied machine learning research is paramount and Africa needs to be at the forefront of that.

I have to say that there is growing interest in this area across the continent and in Rwanda. So last year July, the CMU Africa data science club hosted the first artificial intelligence boot camp in Rwanda for developers in Kigali. Within the space of one week of opening applications, we had over 500 developers sign up. It shows that machine learning is beginning to gain momentum in the community and that’s great for all of us.

Q. What are some of the pertinent issues to harnessing machine learning in the African context and what creative ways do you propose to addressing them?

A. While conversations around building advanced computing infrastructure to enable machine learning and big data are well founded in my opinion the pertinent issue is that we are generating a lot of data as Africans but we are not either collecting it appropriately or storing it appropriately. Fixing that is the crucial first step. The same way you can’t operate refineries without crude oil, applying machine learning and indeed a data-driven economy is based on the premise that representative and accurate data is made available. I agree that we need creative ways to address this issue.

But before we start talking about creative ways, there are obvious things to do that are not being done. Open APIs. APIs provide access to data or unique services by an enterprise. So, for example, the agency in charge of disseminating weather information can provide an open API to allow any developer or enterprise access weather data without having to perhaps manually send email requests.

If you have an enterprise that generates data that could potentially be useful for the public, and yet is not sensitive, you would be doing the economy a service by providing an open API to that data. And there are benefits for your business too. Providing an open API could help you identify new market segments or use cases for your product/service.

Of course, some may say, “we need to digitize first, and we don’t have the money.” Absolutely. We need to be creative. Can we come up with economic models that address the upfront cost of digitization? I think we can.

I’m aware of a group of people in Rwanda who came up with the idea of automating the inventory management of shop owners for free and they basically make money by partnering with third parties that can leverage the data to provide other services. In Nigeria, a startup is rolling out digital solutions for the police force at no upfront cost to the government. But there is a whole economic model around that. Once we start to think about the economic model around the data that enterprise is generating, investing in digitizing starts to make sense.

Q. Today, across the continent young people are the change makers and at the heart of innovative solutions in their countries. What can be done better in view of a pan African innovation ecosystem?

A. Talent, Density, Linking academic and research networks to business, Startup capital.

Talent - For any business, having the right talent is crucial to growth, in the past, companies would choose a location based on things like raw material or cost of labor, but today, especially in technology access to diverse and highly skilled talent pool is key. How does this tie into an innovation ecosystem? Countries need to invest in human capital but also, Immigration is key to any entrepreneurial ecosystem. 50% of silicon valley start-ups have 1 or more immigrants as a key founder. I’m glad that the issues of free movement on the continent are now being fully addressed by our leaders, but one of the things that need to to be done is to create more dynamic labor markets to encourage investment and attract the talent for the ecosystem to thrive.

Density - Innovation is something that is bred through the intersection of great minds. Creating density of talented thinkers and makers dramatically increases the potential for successful ventures to emerge. And that’s why I’m really excited about the Kigali innovation city.  Proximity facilitates the repeated, face-to-face interaction that fosters competition & collaboration and that’s why we need to support cluster growth, create physical hubs and link emerging ecosystems with existing ones.

Linking academic and research networks to business - Entrepreneurs cannot and do not exist in a vacuum; they need to be able to access and build on cutting-edge research and ideas produced by universities and other businesses. Ways to foster strong connections between business and academia include funds for joint research, development of standardized licenses to facilitate technology transfer, and coordination of seed funding for university spin-offs.

Start-up capital: For an innovation ecosystem to thrive, financing is critical, and there’s need to come up with creative ways to attract more venture capital into the ecosystem. An interesting model is the matching model where government co-funds a startup by matching with a certain amount for every investment raised by an entrepreneur.

That’s my take.

By Victor Akinwande

Graduate student in Information Technology, Carnegie Mellon University

victorakinwande.com