AI + Africa = opportunity

14 DECEMBER 2023

Shifting the paradigm – combining existing AI technology with African ingenuity is where our opportunity lies

From OpenAI’s famous ChatGPT to Meta’s Llama, artificial intelligence (AI) models have become increasingly sophisticated and increasingly open, which means they have tremendous potential to revolutionise business. The use cases are endless, and many we have not yet even imagined. But where does the value lie for Africa?

On a continent fraught with basic infrastructure challenges, such as power and lack of connectivity, it is highly unlikely that the ‘next big thing’ in AI technology will happen here. This does not, however, mean that there is no opportunity. At the recently held annual AI in Africa, hosted by Rand Merchant Bank and presented by Arun Varughese, co-head of RMB 's telecommunications, media and technology (TMT) sector, industry experts Michael Jordaan, CEO of Montegray Capital, Nicolaas Viljoen, Technical Lead Director of Artificial Intelligence and High Performance Computing at Meta, Charl Amin, Director of Financial Services for Microsoft Africa, and Chris Erasmus, Country General Manager, South Africa at Amazon Web Services (AWS), gathered to discuss how developments in AI will impact Africa and where opportunities for investment lie.

The evolution of AI

At its most basic level, AI is the art and science of teaching computers to think and act like humans, accessing data and coming up with solutions, sometimes without explicit instruction. While AI is not a new technology, having been around since the early 1900s, it has recently experienced advancement in leaps and bounds, and now many models easily pass the Turing Test, a test of a machine's ability to exhibit intelligent behaviour equivalent to, or indistinguishable from, that of a human.

AI has evolved from big data analytics, to discovering connections within data, to the current generative models, which can produce a variety of content from text to images, audio, and even synthetic data. The potential use cases for AI are endless, and we are already beginning to see new start-ups emerging off the back of generative AI, which would not have been possible without the advancements of the past few years.

Hype, but not just hype

As with any new technology, AI is going through cycles of acceleration, evolution, maturity and acceptance, but it is not all hype. Says Jordaan: “It’s both – there is a lot of hype and many businesses are riding the bandwagon, but it is also very real and is disrupting and changing jobs, technologies and entire industries. AI is constantly evolving as it becomes better trained and has access to more data, and we are now at a profound inflection point for humanity.”

“For us, AI is an enabler for the next version of our company, using generative AI and infusing it through all of our products and technologies. Microsoft is betting it all on AI, from our research and development investments to our choice of partnerships. We believe this is the next paradigm shift, reimagining productivity not just for individuals, but for organisations and companies – and Copilot, our everyday AI companion, is our latest offering to customers in this regard,” Amin agrees.

Copilot is a prime example of AI being used to streamline business processes. The tool is embedded in everyday productivity software including Word, Excel and PowerPoint and uses large language models (LLMs) combined with data in Microsoft 365 apps to boost productivity by creating spreadsheets, presentations, documents and more, and giving businesses powerful insight into their data.   

Getting the basics right

One of the barriers to entry with AI has typically been the cost, but as different models emerge, using different parameters, the cost of using AI is dropping dramatically. As this happens, the potential for new uses cases grows exponentially, with services able to run on much lower cost infrastructure that enables the user ecosystem to become more open and available. However, the basics around data architecture still need to be addressed. AI requires high-quality, clean data to produce effective results, and it is essential that data therefore be captured correctly and stored appropriately so it can be input into AI models.  

Data infrastructure also remains a challenge for South Africa. “Building data centres is expensive and takes time and extensive planning and running them requires significant resources. Networking is critical because latency is a killer for both cloud services and AI. Start-up and scale-up don’t want to have to deal with this, which is where hyperscalers are a crucial piece of the puzzle,” says Erasmus.

Where the opportunities lie for Africa

Africa faces many challenges when it comes to the adoption of AI – not least of which is the fact that access to basic infrastructure is not ubiquitous – and AI uses significant resources, including electricity, which is a resource that cannot be taken for granted in many African countries, South Africa included. We also cannot compete with the massive investment in time, money and skill that have gone into the current AI models, but reinventing the wheel is not necessary to leverage its benefits.

“South Africa is far removed from the ecosystem of expertise and the capital required to develop the primary capabilities for AI. This does not, however, mean that opportunities are lacking. As an investor, I believe the applications of new intellectual property are where the value lies. We cannot afford to let the fourth industrial revolution pass us by, but we need to get the basics right – bringing existing technologies and applying them inside business models in a local context. Data based companies make better decisions and can successfully drive competitive advantage and enhance the customer experience using AI,” says Jordaan.

“As with any new technology, it is essential to look at the business roadmap and start with use cases that will have a notable impact within the organisation, such as productivity gains. For example, infusing generative AI with internal functions such as generating reports, or assisting call centre agents with appropriate information in the moment to improve customer service,” adds Amin.

The key is not to reinvent the wheel, says Viljoen: “It does not make sense for local companies to invest in building their own models. Existing AI models are open source and free to develop on top of – we need to move higher up the value chain and fine tune these models for our own use cases, infusing our own data to add extra value on top of open models. It is about finding the solutions that add demonstrable business value.”

“The pace of innovation is beyond anything we can imagine – the technology is already there, we just need to invest in the skills and use ingenuity to get the applications right. AI will not replace people or businesses, but people and businesses that use AI will replace those that do not,” Erasmus concludes.