Subscribe

SA employees using AI no matter what

Christopher Tredger
By Christopher Tredger, Portals editor
Johannesburg, 21 Jun 2024
Prejlin Naidoo, newly appointed partner at Oliver Wyman’s Johannesburg-based digital, communications, media, and technology practice.
Prejlin Naidoo, newly appointed partner at Oliver Wyman’s Johannesburg-based digital, communications, media, and technology practice.

A staggering 91% of employees in South Africa are willing and able to use AI at work, which means employers have little choice but to better understand the technology and its role in business.

This is according to the results of research conducted by management consulting firm, Oliver Wyman's New-York based think tank, Oliver Wyman Forum, including input from 16 000 white-collar workers globally and 1 000 from South Africa.

The report How Generative AI is Transforming Business and Society is derived from research done in October and November 2023.

Prejlin Naidoo, newly appointed partner at Oliver Wyman’s Johannesburg-based digital, communications, media, and technology practice, says the 91% figure is significantly higher than the global average of 80%.

When asked who uses AI daily in their job, 21% of the participants said they do, Naidoo adds. This then places the onus on employers to regulate the technology, but not block access and stifle innovation.

“For me, that is exciting, but also a little bit scary," says Naidoo. "You could block access to ChatGPT on work laptops, for example, but employees use laptops at home … How do you make sure you provide them with the knowledge and skills to use it effectively? And are they equipped to be able to spot hallucinations and ensure that they keep a human in the loop?”

These are important considerations, particularly because the SA daily AI use statistic is again higher than the global average of 15%, he notes.

Hype-fuelled uncertainty

Hype around GenAI has fuelled uncertainty about the impact the technology will have in many industries. Naidoo says while decision makers are beginning to grasp the  fundamentals of GenAI, they are struggling to answer precisely how this technology will impact their businesses. Use cases are emerging, but doing so randomly.

Naidoo predicts that a few large organisations in Africa may collaborate to build an Africa-specific LLM.

“You don't always need 170 billion parameter model to complete the tasks you need to for a business. Sometimes it makes more sense to train a smaller model that can run with less compute requirement much more efficiently and cost effectively, but still meet the business problem you're trying to solve.”

However, he says the reality for any industry, especially financial services, is that if businesses start using GenAI to make decisions on behalf of customers, at some point regulators are going to insist on a full explanation as to how the business arrived at their decision.

“We've already seen some early moves by the European Union, for example, who have published their first set of regulations. I think their approach makes a lot of sense, which is taking a risk-based approach to regulation. For example, they categorise different types of AI applications based on the risk that they pose to customers, and for the lower risk ones, there's almost no regulation. For the high-risk ones, there's increased regulation. South Africa will probably follow a similar suit, like they did with GDPR.”

Share