Regulating AI in South Africa: an agile approach

Recently, I delivered a keynote speech at AI Expo Africa, Africa’s largest AI event. The topic was “Moving towards regulating AI in South Africa”.

Before and after the speech, I connected with the Expo’s attendees, who had a keen AI interest and came from the private sector, government, and inter-governmental organisations. We agreed that if South Africa wants to leverage AI successfully, we must establish laws that facilitate trustworthy AI.

This post summarises my speech and the lessons I learnt from interacting with the Expo’s attendees. Specifically, this post’s for all stakeholders in the AI industry who want to understand how the country can grab AI-related opportunities while still protecting people from harm. In the end, you’ll know exactly how we use regulation to position South Africa as Africa’s hub for trustworthy AI.

Setting the scene: Africa

The African Union (AU) set a vision for Africa: “A single digital economy”. This vision aligns with the United Nation’s Sustainable Development Goals and the AU’s Agenda 2063.

The question is how South Africa contributes to this digital economy.

South Africa is more than a pretty face

The country has unparalleled beauty. We know this: simply look at the majestic slopes of Table Mountain, the endless seas surrounding the south, east, and west coastlines of South Africa, and the architectural masterpieces of its cities.

But, we’re more than just a pretty face. We have skills, talent, grit, and brilliance. And we want a stronger economy, a level playing field, fair competition, foreign investment, and more. This Grey’s Anatomy meme captures how South Africans feel—Owen is the world; Christina is South Africa.

However, how can AI contribute to moving South Africa forward?

The AI opportunity

South Africa has the opportunity to position itself as the global hub for trustworthy AI. By “trustworthy”, I mean that AI promotes the economy and is not allowed to harm us.

How do we make this opportunity a reality? Through agile strategy, policy, and regulation.

The current state of strategy, policy, & regulation

We need a national AI strategy

We don’t have a national AI strategy. We’re actually lagging behind other African states like Mauritius, Kenya, and Rwanda in this regard.

Instead of overthinking strategy, I suggest government adopts a set of AI principles contextualised within South Africa’s National Development Plan and the Report of the PC4IR. Also, the strategy doesn’t need to be perfect, but it needs to exist.

With strategy comes direction. And with direction, South Africans know where we’re headed.

Next, it’s policy

Once our national AI strategy is in place (sooner rather than later), the Department of Communications and Digital Technologies needs to work rapidly in developing policy that fleshes out the strategy. Key policy considerations include:

• What’s the meaning of trustworthy AI?
• The relationship between AI and data.
• What are the economic interests the country is trying to achieve?
• How do we protect citizens from harm like bias and discrimination?
• Do we need a compulsory national insurance scheme for harm caused by AI?

Regulation AI in South Africa

Last but not least, we need to regulate AI in a way that positions the country as a hub for trustworthy AI. But the tired approach of taking forever to regulate something and then leaving the regulation for 50 years won’t cut it.

We need agile regulation.

An agile approach to regulating AI

I’ve isolated 8 aspects of agile regulation that will move South Africa towards an Artificial Intelligence Act.