DigitalHub

Tech News

How Google is driving artificial intelligence for Africa by Africans

Over the years, artificial intelligence has gone way past the impression of robots in science fiction movies and novels. And its importance has become undoubtedly clear especially in the African continent.

Last year, Google announced plans to launch an artificial intelligence centre in Africa; it finally did in Accra, Ghana.

In this context, the “why Ghana?” question is as important as it seemingly isn’t. When asked, a Google representative said there were no strong factors in play, that it was an effort to evenly distribute Google initiatives across Africa.

Led by Research Scientist, Moustapha Cisse, the Google AI team in Accra is “building AI-powered solutions to real-world problems.”

Moustapha Cisse. Google AI Accra Centre Lead

Cisse who is originally from Senegal is the Lead of the Google AI Accra Centre. He succinctly describes the team and the work they do as;

“A group of researchers working on foundational aspects of AI and applied problems in collaboration with local researchers working in various institutions across the country and Africa.”

The extremely diverse team from 10 different countries are joining forces to tackle fundamental challenges related to AI and Africa.

From education, health to environmental protection and agriculture, the applications of artificial intelligence are endless.

The use of AI to improve human and plant health is very important as the essence of improved food security and life expectancy cannot be overstated.

Using TensorFlow; an open source machine learning library from Google, researchers at Makerere University in Uganda built a program that can detect diseases on cassava plants just by pointing a smartphone camera at their leaves.

Apart from the health of plants and humans, communication is another important aspect the AI team is interested in.

While there are slightly over 250 languages in Europe, Africa has over 2,000 spoken languages making it the most linguistically diverse place on earth.

Cisse said if Africans are able to communicate beyond these language barriers with the billions of people that will live on the continent in the coming decades, it will unlock opportunities and provide access to more information.

All these areas and more are where the Google AI Accra team is working to bring this kind of relevant AI-related research home to Africa and open-source it for people to build great products.

To do this, the team is also heavy on local collaborations.

According to Cisse:

“For AI and related fields, we will provide faculty research grants, PhD scholarships, and sponsor Masters degree programmes; this is already on at the African Institute of Mathematical Science. We are also sponsoring related conferences like the DeepLearning Indaba and Data Learning Africa.”

Democratising AI can be a challenge on many fronts.The world, especially Africa, is fast becoming mobile first in terms of preferred computing medium. This means AI will come largely on smartphones.

Because of price economics, budget smartphones and feature phones with between 256 megabytes to 1 gigabytes of internal memory are quite popular in Africa. In 2018, the International Data Corporation (IDC) revealed that 60% of phones sold in Africa in 2017 were feature phones.

These devices are low performance and resource constrained, so improving accessibility to AI will mean designing accurate and compact learning machines that can be deployed on smaller machines.

The Google AI Accra team says it will be working on redesigning these learning machines to democratise the use of artificial intelligence and making AI more available to many more people.

Being responsibly intelligent

Cisse, himself an advocate of AI for socal good, addresses the global fear of the negative impacts and effects of artificial intelligence; bias and discrimination.

“We do all these research while being thoughtful and careful of the design and use of this technology. At the Google AI centre, we see the work we do as an opportunity to contribute to doing good in the world. This means we look into fundamental questions of fairness and transparency,” he said.

Nyalleng Moorosi, a software engineer and member of the Google Accra AI team whose work is primarily focused on the diversity of AI also touched on the issue.

Nyalleng Moorosi, Software engineer and team member at the Google AI Accra Centre

She said as algorithms are a function of the data they are fed.

“The more diverse sets of data used in building it, the more accurately an algorithm represents the end users it was originally intended for,” Moorosi said while stating the importance of Africans in the process of building AI.

The use cases for AI seem endless and Africa with its growth will be the biggest beneficiary of this boom if properly positioned. And an AI centre for Africans by Africans seems like the best stand.