This guest post was written by Kathleen Siminyu, a data scientist at Africa’s Talking.

Developing countries face many unique challenges. Efforts to address these, in many cases, depend heavily on foreign aid and non-governmental organisations. It is my personal belief, however, that we, the people living in the developing world are the key to saving ourselves.

Artificial intelligence (AI) is the development of computer systems that can perform tasks that normally require human intelligence. Advances in AI have led to its increased use for problem-solving in various industries across the globe. We in the developing world must not be left behind when it comes to crafting solutions to our unique problems.

Regional solutions to regional problems

One such problem is malaria, an infectious disease that, according to the World Health Organisation, killed almost half a million people in 2015, 90% of these in Africa. In 2016, IBM Research Africa hosted a hackathon in Johannesburg to find solutions for anti-malarial drug resistance. The hackathon tasked young scientists to solve challenges that would have a significant positive impact on the continent.

Equipped with malaria genomic datasets, the scientists developed an artificial intelligence solution capable of determining baseline models for predicting drug sensitivity and malaria drug combinations. They also attempted to predict which malaria isolates would be resistant or sensitive to artemisinin, the most effective anti-malarial drug today, whose effectiveness is threatened by emerging resistance.

The success of the hackathon shows what can be achieved when existing commitment and skills are channelled to solve some of the region’s biggest problems.

Beating traffic

Another challenge in many developing countries is chronic traffic congestion. A World Bank study found that in Uganda, the problem costs the country over US$800m in lost GDP each year. Rapid urbanisation without sufficient expansion of roads is compounded by a lack of real-time monitoring and management. High costs often mean the deployment of conventional traffic flow monitoring systems in the developing world is unfeasible. In addition, these systems often hold assumptions inappropriate to traffic flow in these countries, for instance, that vehicles remain in dedicated lanes.

Rose Nakibuule and John Quinn from Makerere University have designed a low-cost vision-based system as an alternative, suitable for the developing world. Mobile phone cameras are used for data capture and transmission, with the phones enclosed in steel boxes with solar panels for charging. The cameras capture and upload images to a server through a wireless internet connection.

Once images have been uploaded, the correspondence between consecutive pairs of images is calculated so as to obtain flow vectors corresponding to every moving object. Road geometry is then inferred in relation to the camera to project those flow vectors into ‘real world’ coordinates, allowing the software to calculate speeds of moving objects and thus exhibit intelligent behaviour by predicting traffic flow.

Evaluation of the prototype shows that it is not only less costly to deploy than video systems currently on the market, it also required little maintenance despite being fully exposed to tropical weather.

Building capacity for AI research

These are just two of the many examples of researchers using AI to build local solutions to local problems. If we focus on further developing local skills, we can have a far greater impact. Currently, few institutions of higher education in sub-Saharan Africa have the faculty to support students carrying out research in AI. I know of only two institutions offering courses within the field – the University of Witwatersrand and the University of Pretoria.  And then there is the very real funding problem – the dearth of basic research funding for African scientists at African institutions solving African problems.

We need individuals with an inherent understanding of the challenges that AI can address, with the knowledge, skills and support to design and implement these solutions for our context. We need AI-powered businesses that are well-funded and have successful business models to provide the infrastructure for technological advancement in this field.

We need to make sure that Africans are not just recipients of advances in artificial intelligence but shapers and champions as well.

For more insights, follow Kathleen on Twitter at @siminyu_kat and the Web Foundation at @webfoundation

The Web Foundation recently published a series of white papers looking at the impacts of AI, algorithms and control of personal data in low and middle-income countries.