Africa benefits from an incredibly young, tech savvy population; over 60% of the population is under 25 years old and in just Senegal the average age of the population is a mere 19 years old. That makes it a hotbed for tech development, which is already flourishing, but it still has a lot of work to do.

“Much of tech development matures over a two decade period. That means we don't know what some 65% of future jobs will be and the majority of people that will be using this technology are not even born yet,” Dr Moustapha Cisse, the CEO of Senegal-based Kera Health told delegates at the European Bank for Reconstruction and Development (EBRD) annual meeting in London on May 15.

Africa faces a lot of structural problems. Over two thirds of businesses in Senegal complain that they face regular blackouts. However, mobile phone penetration is high and mobile payments are widespread creating a foundation for e-commerce and online business.

And Africa needs talent. The government has not invested heavily into technical education, but there are some private sector colleges backed by Meta and Google that have produced hundreds of engineers.

“If you have used Google Maps then part of the algorithms in that app were built by African engineers here in Africa,” says Cisse. “But we need to invest more. People are the first pillar of the technology infrastructure.”

The panellist also underscored the cultural bias inherent in importing AIs developed in Silicon Valley and elsewhere.

“The development is based on the infrastructure that has been built up over decades and it's impossible to separate out the cultural background that forms the context for this development,” says Cisse. “We need to train African engineers so that we can develop software that is relevant to the African cultural heritage that is very different to that in the rest of the world. There should not be one AI to rule them all.”

No one denies the potential of the African market. Currently in the top ten of the world’s fastest growing markets, just the digital payments market is expected to top $1.5 trillion a year by 2030, but development remains at a very early stage.

Invest

And relatively small investments can be transformative. Access to funding remains an issue, but relatively small amounts can have large effects.

Bigger investments could also develop a whole industry. Matthew Blakemore, CEO of AI Caramba! pointed out that Africa is home to the whole periodic table of critical and rare earth metals but typically exports them for processing in China.

“But Africa would develop its own chip industry based on these materials with suitable investment,” said Blakemore. “The same is true of data centres: Africa is home to 17% of the world’s population, but has only 1% of the data centres.”

The EBRD showcased Armenia’s experience as a case study of what could be done. Companies there have leapfrogged the normal development chain and went straight to state-of-the-art. With some $75mn of funding, assisted by the EBRD, the telecoms sector was modernised and 270,000 homes were hooked up to fast internet, turning the country into a digital hub in the Caucasus.

Tower of Babel

The tech entrepreneurs on the EBRD panel on African tech all pointed out that the technology needs to be developed in an African context to work best; that simply importing western technology is a hindrance to tech development in Africa.

Africa has 54 countries and over 2,000 languages; the language diversity inside Africa is greater than in all the countries outside of Africa.

If AI is developed on an English platform and imported and adapted to Africa and its polyglot language base then there are higher costs and that is more expensive to use than an AI trained on multiple local languages and then adapted, says Dr Cisse.

“It’s basically a tax on your inability to speak English,” says Cisse. “And necessarily it contains the world biases of the countries where the AI was developed.”

To illustrate, Cisse explained that the “social media” in Africa is not text based messages sent by smartphones, but the radio.

“Everyone listens to the radio to get their news and entertainment,” says Cisse. “That’s our social media. That data can be leveraged and used as the data for AI, but it is more complicated and more expensive to do than with text based data.”

The EBRD also showcased tech projects in Kyrgyzstan where the Akylai AI was built on the local Kyrgyz language. It learnt 300,000 words of the language and has 15,000 users to be able to give its answers in Kyrgyz as part of the EBRD’s Star Ventures programme.

“They made global local and the model can be replicated,” Nancy Kacungira, a BBC correspondent and the panel moderator said.

Other models that focus on putting indigenous languages online have attracted the attention of mobile phone companies in Africa, who are looking for ways to get closer to their companies; the appeal of these AIs is that mobile phone companies can offer an interface in multiple local tribal languages.

Heiura Itae-Tetaa set up E-Reo that was a finalist in the recent Aurora Women in Tech global awards. From the tiny French Polynesia island of Tahiti, her platform preserves indigenous languages that are dying out. 

“These languages are not present in the digitalised world and are slowly dying out. We have already captured 50 languages and hope to have 400 by next year,” Itae-Tetaa said. 

Government’s and multinationals have supported the programme for cultural reasons, but mobile phone companies are also customers as E-Reo allows them to present their services in the local language; in Nigeria there are half a dozen dominate languages but the country has over 200 different languages in total and E-Reo is including them one by one.