As AI takes off globally, Africa has struggled to keep up. The challenges contributing to this struggle are aplenty but can be summed up into three main troves: lack of data/infrastructure, and capital funding. Addressing these challenges is key to bringing Africa into the AI foray, and it is what the BRAIN Bootcamp sought to help its cohort startups achieve.
When OpenAI launched ChatGPT in November 2022, the Large Language Model (LLM)’s promise was to put all the internet’s information in one chatbot. ChatGPT’s power was in the datasets, which were used to train the model to give accurate responses. However, the challenge of ChatGPT’s inaccuracy in resolving queries associated with Africa was apparent because of the lack of datasets to train the LLMs.
The key to resolving the lack of reliable datasets challenge in Africa to train LLMs can be effectively addressed by small language models (SLMs), according to Younes Bensouda Mourri, a Stanford University lecturer of Computer Science.
“SLMs are key to accelerating AI solutions in Africa because they require less training infrastructure, which is very expensive to get,” said Mourri. African startups working on AI solutions are better off utilising SLMs, which work with specific and, most importantly, available datasets.
Mourri was speaking at the recent BRAIN 4.0 Bootcamp in South Africa, hosted by Open Startup (OST), powered by AfricaGrow, Digital Africa, and Sanofi’s Global Health Unit for sustainable impact, run in partnership with MIT-Africa, the MIT Executive MBA program, AfricInvest, InstaDeep, and Bpifrance. This regional bootcamp was primarily hosted by Stellenbosch University, LaunchLab, CERI, and Fluorobiotech, with sponsorship from Octoco, Technopark, and Workshack, and local support from Wesgro and Innovation City.
To address the capital funding challenge, investors like Digital Africa, who have been sponsoring the program for three years, have been backing AI startups on the continent even before AI became mainstream. According to Malek Lagha, Acceleration Manager at Digital Africa, programs like BRAIN are vital in scaling up deeptech innovation in Africa.
“One of the most difficult things for startups in deep tech is transforming their technology and hard science into a solid commercial undertaking”, says Lagha. “This is exactly where the BRAIN bootcamp comes in, allowing the experts who are present here to help founders address these challenges.”
The BRAIN 4.0 cohort comprised five startups that are working on AI solutions across different sectors in Africa.
“As Digital Africa, we trust partners who support startups on the continent like OST. Programs like BRAIN are vital and need financing and knowledge from other partners, because partnership is really important to go to other countries in Africa, and we are glad to play a part.”
Through presentations and one-on-one sessions held over five days, experts drilled into the cohort startups on how to differentiate their AI solutions. The sessions culminated in pitch sessions where experts, including representatives from VC firms Agrivest and Sanofi, had the chance to assess the startups’ progress.
TanzMed, a Tanzania healthtech startup, was one of the five startups in the bootcamp. The company is digitising healthcare services to boost efficiency in a country where the doctor-to-patient ratio is 1:20,000.TanzMED has over 300,000 individual customers who also contribute the data needed to train the model. According to co-founder and CEO Mkata Nyoni, SMLs have been crucial in enabling the company to scale its solution.
“SLMs allow us to train our systems so that they can make a lot of decisions with minimal data, which is relevant to us because existing LLMs do not have sufficient data on Tanzania, so we use our datasets to train the SLMs,” said Nyoni.
As artificial intelligence takes charge and looks set to upend several industries, Africa mustn’t be left behind. This is where programs like the BRAIN 4.0 Bootcamp need to guide African startups on how to take advantage of fringe AI concepts like SLMs, which are more relevant and practical for the continent.
A combination of expertise from experts like Younes and Richard Rabbat, CEO of Lighty AI and the support of long term funders like Digital Africa and Sanofi is key to bringing Africa into the fray.
