It is estimated that Gen AI will have a market value of $60-billion by 2025 and account for about a third of the total addressable market for AI in general. In the past three years, Gen AI startups attracted about $20-billion in investments.

While AI refers to machines that can perform tasks that normally require human intelligence (such as reasoning, learning or creativity), Gen AI refers to a subset of machine learning techniques and models that can produce original text, graphics or audio. ChatGPT is already known (a text-based AI) and so is Dall-E (text to image AI). The take-up has been rapid: ChatGPT reached a million users in five days.

Specific sectors where Gen AI can help SA

Our recent report, developed in collaboration with Microsoft and Wits Business School explores the potential impact of AI and Gen AI across four key sectors of our economy. South Africa’s healthcare sector is grappling with unequal access to care and affordable medicines, variable quality in the care provided, inadequate funding and a shortage of staff, particularly in the public sector. The doctor-patient ratio in 2019 was 0.8 doctors per 1,000 people, compared with 3.0 in the UK and 2.3 in Brazil.

Read more in Daily Maverick: Health in 2023 highlights — SA healthcare has endured a critically busy year

Gen AI tools can transcribe and summarise each consultation and maintain patient records, freeing time for doctors and nurses to treat patients. An AI tool called DeepScribe is being used for this purpose by medical practitioners in the US. AI also frees up professionals’ time by supporting personalised diagnoses and treatment regimes for patients by analysing data and taking account of available inventory in hospital pharmacies. However, doctors will need to maintain a reasonable level of distrust to protect against AI hallucinations and data bias.

As in health, there are severe disparities in access to good education in South Africa. Poorer schools suffer from a shortage of teachers, who often lack the necessary resources. This is obvious in results: a recent report found that 81% of Grade 4 learners are still unable to read for meaning.

Read more in Daily Maverick: Call it what it is — the SA education system is in complete ruins

By analysing vast amounts of data, Gen AI can inform policy and curriculum design and create learning materials. Content can be tailored to various grade levels and individual students and presented in the student’s preferred language. In France, NolejAI enables learners to upload text, video or audio, and creates interactive lessons. 

Financial inclusion is another area of inequality. About 20% of South African adults lack even a basic bank account. Most are women, people living in rural areas, and from low-income households. Expanding financial inclusion would raise levels of financial literacy, money management and access to opportunities.

Read more in Daily Maverick: Digital advances set to drive financial inclusion in SA’s informal sector – Payfast report

Gen AI can change the way banks interact with customers. AI-powered chatbots can handle routine queries and provide timely responses, which spares customers the effort of visiting branches and increases the transparency of, and trust in, products and services. Some financial institutions are using Ushur as an engagement platform that uses conversational AI to provide streamlined digital interactions for customers and data to know customers better. It can deliver almost a tenfold faster response time for customers.

In other areas, Gen AI can provide personalised recommendations, education and financial advice and help draft legal documents and explain, in plain or home languages, their pertinent terms.

Agriculture contributes about 2.5% of South Africa’s GDP and about 10% of the country’s exports. But farmers face droughts and climate change, as well as difficulties accessing loans, land tenure and markets.

Read more in Daily Maverick: South Africa’s small-scale farmers have a hard row to hoe, especially in the absence of policy support

Gen AI can analyse data, helping farmers to follow more efficient and sustainable farming practices. Using sensors, drones and satellites, farmers can accumulate data on their crops, which can be analysed by AI algorithms. The results will guide farmers on how best to use resources and maximise crop yields. An integrated AI platform, Jiva, helps smallholders with an advisory engine, enriched by machine learning, that provides advice on sustainable farming. It uses climate data, remote sensing and computer vision to assist farmers in decision-making. Reportedly, farmers using Jiva have increased crop yields by up to 40%.

Other applications of Gen AI in farming are to monitor crops for early signs of disease or pest infestations, enabling early intervention. With refined credit scoring, AI can expand financial access for smallholder farmers. That will empower smaller farmers to invest more in their operations, adopt modern agricultural practices and improve their standard of living.

Managing the risks of Gen AI

For South Africa to realise the full benefits of Gen AI, the correct legislative framework will have to be put in place to manage the risks, quickly, to match the speed of Gen AI roll-out.

One of the risks of Gen AI is job losses. South Africa had an unemployment rate of 33% in the second quarter of 2023. However, some jobs will be displaced, not lost, as Gen AI could open up new work opportunities. Employers could be incentivised to retain workers and/or upskill them. AI-focused businesses, particularly start-ups, could be supported with legislation. Through public-private partnerships (PPPs), the public and private sectors can pool resources to roll out AI. One of the benefits of this collaboration would be to ensure that AI solutions create advantages for everyone.

Other important participants in South Africa’s Gen AI roll-out are Development Finance Institutions (DFIs) and venture capitalists, who can support businesses and research institutions. PPPs and DFIs can also assist in financing AI infrastructure. Data-sharing models are needed to build the large, diverse datasets on which AI models are trained, and at a local level, stable and high-bandwidth connectivity has to be available full-time.

Once all the correct building blocks are in place, Gen AI will be positioned to play a transformative beneficial role across key aspects of South African society. DM