From dance to diagnosis: How Tensorflow.js is shaping AI in Africa

Description

As the western world continues to explore advancements in artificial intelligence, sub-Saharan Africa lags behind. This is in part due to low internet connectivity, lack of proper information and communications technology (ICT) policy implementation, and a stunted region-wide open data economy. These factors combined greatly limit the application of common macro-AI instances that rely on big data for success.

In the face of these challenges, there presents the opportunity to harness unconventional proxies for datasets that could be used in lieu of a preferred value set. An example of this is how a United Nations agency Babusi Nyoni consults for, used the price of a goat in Somalia through collaboration with displaced populations as a proxy for displacement signals. This was after it was discovered that the market price of a goat could decrease relative to the number of people whose departure from an area was imminent.

Babusi dives into opportunities presented by small data in the development of AI-enabled solutions for the African continent. He explains how mobile-first internet connectivity trends in the sub-Saharan region have readied a generation for more personal relationships with technology and how Triple Black’s dance app experiment showed that there’s a genuine willingness on the continent to interact with new technology as long as it’s relevant. He also unpacks the approach used to apply the dance app algorithm to build a prototype for the early diagnosis of Parkinson’s disease and how, in underdeveloped communities, client-side AI such as TensorFlow.js is the future.