The current state of AI and ML in Africa

Recently at AI EXPO AFRICA 2020, I was interviewed by Dr Nick Bradshaw, where we had a quick chat on the current state of AI in Africa. I shared my thoughts, in the interview, on the progress that I’ve seen over the last 2-3 years , but also some of the challenges facing us in the future. These challenges include :

• Education – while AI will definitely have an impact on certain types of jobs in the future, corporations and governments need to make education accessible ( and cheaper ) to move people into new types of roles that will emerge. Countries that do this will enable their populace to progress up the value chain– countries that don’t will suffer tremendously. AI does not have to decimate the job market , but certain types of jobs will be at risk, and I will explore the impact of AI to jobs from an economic perspective in an upcoming article.
• Use cases – Until recently, I've found that data scientists were struggling to find business buy in for Machine Learning use cases. While this has improved , I proposed the Microsoft “4 pillars” of digital transformation as a guideline for developing these use cases, for maximum business impact.

When you focus on specific problem areas , like "How do we improve the customer experience?" , or "How do we make internal operations more efficient?", you will develop use cases for ML that are measurable in their value to business, and thus probably see better interest from business.

Lastly , I also mentioned that the penetration of AI will continue to happen very swiftly and transparently. I spoke of how microcontrollers / CPUs slowly spread across equipment / cars / appliances in the 20th century, bringing simple logic to those devices and improving operations. Consider when cars moved to electronic engine management systems – you now could build in simple IF-THEN-ELSE decisions into software that was running the engine , leading to more efficient operation.

The auto-enthusiast knew of exciting development and could see the difference, while the average person simply noticed that his/her newer model car worked better ( perhaps starting easier on a cold morning ).

Now, we see ML models being deployed everywhere , assisted by ML capable hardware ( ie. neural network engines in CPUs for example ). An example that I used was my phone unlocking by seeing my face – there is a neural network chip onboard , which is running software with a trained model on it. This progress happened seamlessly, all most consumers knew was that their new model phone could do this without knowing all of the above. And so it will continue……

The full interview is below :