The African Approach to Artificial Intelligence

The Vice- Chancellor & Principal of the University of Johannesburg, Prof Tshilidzi Marwala recently penned an opinion article published by Forbes Africa, November Issue.

Despite the political and popular confusion, it is common cause that we are no longer on the brink of, but live in the era of the fourth industrial revolution (4IR). The main driver of 4IR is artificial intelligence (AI). According to President Vladimir Putin of Russia, any nation that masters AI will dominate the future. AI is a technology that makes machines “intelligent”. Traditionally, we considered intelligence to be an attribute limited to biological organisms, such as humans. A machine is considered to be intelligent if it can analyse information and extract insights beyond the obvious. The idea behind AI is to make machines intelligent, and in many tasks making them more intelligent than humans. For example, an AI machine designed by IBM called Deep Blue plays better chess than any human while Alpha Go, designed by Google Deepmind, plays the game Chinese Go better than any human.

AI is transforming all aspects of our lives. An AI system designed by Cambridge Analytica, for instance, is alleged to have influenced the presidential election in the United States in favour of President Donald Trump. Uber is in the process of launching a self-driving car, even after one of its cars killed a pedestrian during the testing stage. AI systems are now as competent in reading medical images as expert radiologists.

Given the centrality of AI in the modern era, how do we capacitate our people to understand it? One way of doing this is to organise intensive short courses on AI. An initiative such as the Deep Learning Indaba, the brainchild of Dr Shakir Mohamed from Google Deepmind, aims to strengthen machine learning in Africa by organising meetings (called Indaba in isiZulu) around the continent and getting world experts to come and teach about AI. Incidentally, Dr Mohamed was my Master’s student at the University of the Witwatersrand. The Deep Learning Indaba on average attracts people from more than 30 African countries and 19 from other parts of the world and has so far held AI meetings at universities in South Africa, Kenya, Rwanda, Ghana, Nigeria, Morocco, and Senegal.

Initiatives such as the Deep Learning Indaba are aimed at people who already have tertiary education. However, one of the biggest concerns is whether we can teach children concepts that would facilitate easy learning of AI later in their lives. Fernando Buarque from the University of Pernambuco in Brazil, as well as Nickey Roberts, from the University of Johannesburg, and I have just published a book on AI for children titled “My First AI Book- Artificial Intelligence and Learning.” This is the first book of its kind and will form a series of six books on AI. Using cartoons and elementary examples, the book breaks down AI concepts such as methods of machine learning, coding, and human-machine interface.

However, to make this book relevant to Africa, certain things still need to be put into place. The first of which is the translation of the book into local languages. We are translating this book into languages such as isiZulu, Swahili, Tshivenda, Kinyarwanda, and Yoruba. However, translation into local languages is not as easy as it sounds, as many of the AI concepts are not available in our local languages. As part of the translation process, we have to invent words which expands the vocabulary of our local languages.

The second dilemma that we faced in translating the book is that some examples that we use are not universally understandable. An example of a young boy skating to illustrate the concept of reinforcement learning is difficult to explain. Reinforcement learning is the concept of learning through the principles of punishment and reward, but skating is not generally recognised in the African continent, especially in rural areas. We have found that translation in the African context is more than just the translation of words but also the translation of context, which can be difficult.

The third dilemma that we faced was how do we use different platforms to facilitate the learning of AI. In this case, we chose short movies to illustrate storylines and AI concepts. However, the disadvantage of using movies to illustrate concepts is that many of these movies are transmitted using the internet, which is difficult to achieve on parts of the African continent because of limited internet connectivity.

Part of the solution should be introducing 4IR technologies such as AI into our schooling system, especially in the first few years of learning and for early childhood development. Secondly, African countries should invest in translating AI technologies and concepts into our local languages to enable understanding. Thirdly, where possible, African countries should tap into different media platforms such as the internet to aid the teaching of concepts not commonly understood such as AI.

  • The views expressed in this article are that of the author/s and do not necessarily reflect that of the University of Johannesburg.

Related Articles

READ: Artificial Intelligence: An inhumane future?

READ: UJ’s Prof Tshilidzi Marwala sheds light on data privacy in the era of AI

READ: UJ’s Prof Tshilidzi Marwala explores whether AI can make us more rational

prof tshilidzi marwala
Prof Tshilidzi Marwala Vice- Chancellor & Principal of the University of Johannesburg
Share this

Latest News

All News