Artificial Intelligence in Low/Middle Income Countries; The East African Experience

I recently received a request to be a speaker at an event that the Web Foundation, the Center for Global Development and Future Advocacy were hosting to discuss the implications of their recent work on the economic, social and political impacts of artificial intelligence in low/middle income countries. I wasn’t able to attend the event, held in London, in person but joined in via video conference and this was my little contribution: AI in low/middle income countries, the East African experience.

As a practitioner in the field of AI who works in Kenya, I come across exciting and innovative ideas and applications of Data Science and Artificial Intelligence. More satisfying than just reading about them is meeting and interacting with people who are using AI to locally solve problems in Kenya specifically and the wider African continent in general.

In this commentary, I will highlight companies and innovations in various sectors, starting with the more ‘mature and established’ ones to those that are still in the conception or research phase. What’s described below tells the story of the AI landscape in Kenya and East Africa and how it is heavily dictated by the problems that ail us.

Kicking off with money matters, a Pew Research Centre survey done in 2015 showed, “the percentage of people living on less than $1.25 a day in sub-Saharan Africa is more than twice as high as any other region in the world.” With so many people living on so little, most of the population in sub-Saharan Africa is excluded when it comes to traditional financial services.
In 2007, the leading mobile network operator in Kenya, Safaricom, introduced mobile money – a product known as M-pesa – and disrupted financial services in Kenya. Ten years later, with approximately 27 million subscribers, Safaricom remains a leader in the mobile banking sector, having introduced other products over time such as M-shwari, which is a banking product for M-pesa users that allows you to save and borrow money through your phone while earning interest on money saved. Using information such as saving and spending patterns, they determine how much of a loan you are entitled to thus providing credit to a whole spectrum of the population who would otherwise not have access to it.

Other players in the space include Tala, a company that provides micro loans in several countries world wide, those in Africa being Kenya and Tanzania. They offer a smartphone application that collects data ranging from biographical information, the number of people the loan applicant contacts daily, the size of the applicants’ network and support system, their movements, routine habits like whether they call their mother everyday or pay their bills on time. They find these to be more meaningful towards credit scoring.

Easing into agriculture, a report released by One Acre Fund on scaling up agricultural credit in Africa stated, “Of the more than 1 billion people in the world who survive on less than $1.25 per day, 75 percent depend on agriculture for their livelihood.” Small-scale farmers, contributing a large percentage to those living below the poverty line, do not have access to traditional banking services. There are institutions specialized in providing these to them.

Using machine learning to process alternative data, not necessarily pertaining to the actual crop output, such as the farmer’s individual and social data, environmental and satellite data these companies can determine how much to lend farmers based on their expected agricultural output. Examples of companies doing this are FarmDrive and Juhudi Kilimo.

Additionally, there is research using data collected using IoT devices. An IoT device is any nonstandard computing device that connects wirelessly to a network and has the ability to transmit data. Research using these is not widespread due to the added cost they present. One project in Tanzania uses sensors to predict when cows are on heat for optimum insemination timing, because semen is expensive.

What compounds the problem of collecting data using IoT devices is data storage. This data could be easily streamed to the cloud but Internet connectivity is not always reliable in most places in Sub-Saharan Africa, especially in rural areas. An alternative is to cache the data locally, on the IoT device itself, but the downside of that is the requirement for the manual retrieval of the data. In cases where the data is time sensitive and a decision needs to be made as soon as possible, the delay may hurt the decision making.

KUDU – a mobile based agricultural market in Uganda – started out as research looking into how mobile phones could make agricultural markets more effective. They found that farmers had trouble finding buyers for their produce and were facing the threat of spoilage of their goods while on the flip-side traders faced uncertainty when it came to locating produce. To solve this problem, they developed a double auction system where buyers and sellers submit their information separately and the best matches are found computationally. Another project from Uganda uses smartphones to capture images of crops and diagnose disease with computer vision techniques. These two projects are under the AI Research Lab in Makerere.

Urban planning is another sector where data analysis is making a difference.
Nairobi ranks among the most congested cities in the world. It is said to lose US$357m annually in fuel consumption, pollution and lost productivity due to its notoriously endless traffic jams. How to beat its traffic problems is an ongoing, heated debate.

Towards this end, Ma3Route built a community on its platform where traffic data is crowd sourced and the information shared is trusted. Using simple text and photographs, mainly posted on Twitter as well as their mobile application, users share their joys and frustrations on the road thus helping others decide whether or not to use a certain route.

Due to the large amount of information collected on where people go at what time of day as well as how they get there, they have several data-backed projects such as an accident mapping project that collects accident reports from users and plans to release a comprehensive report. Such reports could then be used to shape policy decisions around road safety.

In Uganda, a company called Thin Void started out trying to solve the problem of bodaboda theft. Bodaboda are bicycle and motorcycle taxis commonly found in East Africa. Thin Void attach GPS sensors on the bikes, which collect data on where the bodaboda are as well as how fast they are moving. This data is transmitted every 30 seconds. Over time, they came to realise how valuable the data they collect is and started exploring it for other uses such as traffic congestion monitoring.

Among other interesting projects, IBM Research in Kenya tries to address the problem of energy access and reliability. When it comes to energy access, one of the main challenges is infrastructure planning. Some areas, such as the cities have a high population, while others especially in the rural areas are very sparsely populated. As a result of that, infrastructure costs might be very high without getting justifiable returns when it comes to expanding the power grid to rural areas. Attempting to predict electricity demand for new customers helps providers know how much to generate and improves reliability by reducing the chances of outages due to overloading of power systems.

In education, one player shaking up the space in Kenya is Eneza Education who use low cost mobile technology to give users educational lessons and assessments through SMS, web and android platforms. As of 2016, Eneza has over 55,000 monthly active subscribers, over 1 million lessons viewed and over 10 millions question answered. All this translates into a lot of data which they are starting to explore and disseminate with the aim of giving parents and schools insights on how to best help their students.

Still in the research phase, there is some work at a local university going into predictive modeling of academic performance whose aim is to identify factors that contribute to poor performance in school with the objective of placing preventive measures to reduce failure rates.

In health-care, Savannah Informatics in Kenya tries to solve the problem of care fragmentation by working with partners to deliver an integrated health information system. As things stand, most health records are paper-based, and thus confined to the health centre/hospital where they are collected. Savannah Informatics is working on a centralized and integrated system. Using this data and in collaboration with their stakeholders, they also help in reviewing, developing and implementing medical curricula.

Brave Venture Labs, specializing in algorithmic human resources, aims to unleash the African potential by linking people to opportunities. They do this using a combination of individuals’ social media data, African based psychometric measures and inexpensive short learning programs which they offer.
There are fields that I haven’t had a chance to go into but allow me to conclude, Artificial Intelligence despite still being very ‘young’ in East Africa, is very vibrant. Not many people have specialized training in the field but there are communities that bring together individuals who are self teaching, be they academics, industry professionals, students, basically people from all walks of life. This is creating room for collaboration and further exploration of solutions to the problems we face.
Personally, I am excited for the future.

Like it? Share it!Share on FacebookShare on Google+Tweet about this on TwitterShare on LinkedIn

1 Comment

Add Yours
  1. 1
    Sophia Nasimiyu

    I love this. so informative.
    I can’t wait to see what tech in East Africa, specifically Kenya, has in store for us in the future. ?

Leave a Reply

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