Sun. Jul 14th, 2024
    artificial intelligence
    artificial intelligence
    "​...but we are in Zimbabwe..."​: Contextualising the Artificial Intelligence (A.I.) hype 6

    Two weeks ago, I had the opportunity to present at Steward Bank in Harare, Zimbabwe. 

    1541396255843?e=1668038400&v=beta&t=vXZAtst2m1t2c3BhTBd18RN3scikDzkNJ5GdD QnN7M

    The discussion was about Artificial Intelligence, Machine Learning and Deep Learning. I will go through the whole presentation in upcoming posts. For now, let’s focus on a section where I discussed some interesting use cases. The developed world is awash with examples where these technologies have been successfully implemented but I was presenting to an analytics team based in Zimbabwe, so the expected comment from the team was, “…but we are in Zimbabwe…” This is the reason for this post (and some others to come because I think contextualizing the hype around these technologies is fundamental to their implementation on our continent). I hope these posts will show that even us in the the developing world have much to gain by seriously getting into the game as well.

    Uses of A.I. in developing countries

    There are some problems which are more prevalent in developing countries than the first world. This means there is a lot of catching up we have to do. Technologies like A.I. can help us leapfrog some of those steps in important industries/sectors if we are to develop Africa. Before I get into what we stand to gain, I do concede the fact that these technologies are disruptive for jobs many of us are employed in. For example, in the Philippines, call centers employ 1.2m people and account for about 8% of national income. A.I. technologies are advancing at such a pace that soon we will be interacting with automated agents instead of warm bodies when we call help centers (cue Google Duplex). What our governments need to weigh is whether these disruptions are worth the positives we will get in return.

    Use case 1: Education


    One major problem African countries face is the lack of experienced teachers. This is huge issue even in well-resourced countries like South Africa. Tools are being developed that help teachers track how users interact with the content, be it videos, practice questions etc, and the software automatically creates a learning plan for the student. These A.I. tools personalise education – they create learning paths which are tailored to each child’s strengths and weaknesses. If a child struggles with 2 + 2, maybe it is because they have not understood 1 + 1 yet and these tools are able to pick this up and direct the child to the start with the right sections. One might think this requires some fancy technologies only the first world countries can afford but it turns out, all these very cheap Android phones coming out of China into Africa are very much capable of being educational tools when apps are coupled with the right A.I. algorithms. Some places in Africa are still too remote and students have to walk many kilometers to get to school. The South African government gave an ultimatum to service providers Vodacom, MTN and Cell C that each network’s services has to cover the whole country by 2020, either by improving their services or opening their networks to MVNOs. It is not difficult to see how a combination of improving mobile network services (and hopefully cheaper data plans) and increased influx of cheap and capable/powerful smartphones will make it possible to build the right apps with A.I. backed capabilities that will help us deliver the best education even in the remotest parts of Africa. Even without network connectivity, Google’s Tensorflow Lite, is being developed to allow for “on mobile device machine learning” without the need to connect to the cloud. Very useful capability for remote areas, don’t you agree?

    Use case 2: Healthcare

    1541437510729?e=1668038400&v=beta&t=3OXzqkCb 8vJ9pdK6D b1eVb7RerAYOvFFo g7XY OY

    The remoteness of some African villages presents some very complex healthcare challenges. It is sometimes difficult to deliver drugs to areas where they are needed the most. This is where A.I. backed technologies can be useful. In Rwanda, a startup called Zipline is delivering medical supplies using drones and they claim they have reduced waste of donated blood by 95%! Some A.I. algorithms predict the spread of outbreaks such as Ebola, giving healthcare practitioners ways to prepare counter measures. Other tools are allowing nurses and lesser-trained medical practitioners to diagnose patients in remote areas through the use of A.I. backed mobile apps. One example is Retina.AI by Nigerian doctor Dr. Stephen Odaibo. It is a mobile app being developed with the aim of “using artificial intelligence to improve retina care.” All these A.I. backed medical tools allow us to diagnose medical conditions at a fraction of the cost, which is definitely beneficial to emerging markets. There is a promising algorithm being used to diagnose Malaria, one of Africa’s deadliest diseases, at low cost and higher accuracy than standard methods. Although not developed on the continent, is a mobile app that aims to provide 24/7 care to people with depression, stress, and anxiety. “Built using state-of-the-art predictive models developed by MIT scientists and engineers, uses a combination of smartphone technology and data analytics to deliver personalized and affordable healthcare.” Won’t Africa and emerging markets benefit from these A.I. backed technologies, considering the challenges we face which make healthcare delivery expensive if not impossible?

    Use case 3: Agriculture


    Many African countries are dependent on agriculture; Zimbabwe, where I did the presentation, was once called the “bread basket of Africa”. But this activity is dependent on many factors including access to capital, right climate, population dynamics, labour, and changes in soil composition. There are A.I. systems being developed to maximise each of these resources. Predictive models have been developed to assess the impact of environmental factors on the yield of different crops, thereby allowing the farmer to plant just the right type of crop to maximise his or her return. Startups are making use of drones and computer vision to keep track of how crops are growing by taking pictures and analysing the size of the leaves, and heights of the crops so as to deploy remedial resources in sections where crops are failing, again maximising yield. A company called Blue River has developed a system that employs computer vision where, if their “See and Spray” robot sees weeds, it automatically sprays herbicides. This robot can work 24/7 with no need for human intervention, allowing the farmer to shift his labour resources from tilling the land to other more productive aspects of farming. In due time, I will write a post about a startup I am researching which employs some machine learning in agriculture for anyone to implement (stay tuned!). With Africa’s dependence on agriculture, I have no doubt these A.I. backed technologies will definitely benefit our economies.


    For this post not to be a long read, I will write about other industries in upcoming posts, for example making use of A.I. backed solutions to bank the millions of unbanked Africans. One very important point though is, if there is this much to gain from seriously looking at these technologies, what are our governments doing? The UAE created the world’s first post for the Minister of Artificial Intelligence. President Putin declared that whoever controls A.I. will rule the world. Yes, these countries are more advanced than we are but they give direction to what will come towards us as we try and grow our economies to their level, save the fact that they dictate the direction the world goes in. Are we preparing ourselves enough to leverage these technologies or are we going to be left in the dust again thinking this revolution is only applicable in first world countries?

    61 / 100

    Leave a Reply

    Your email address will not be published. Required fields are marked *

    This site uses Akismet to reduce spam. Learn how your comment data is processed.