Kgalema Motlanthe is a highly respected elder statesman in South Africa. He was President of the country some 15 years ago. Even though he belongs to yesterday’s generation, he is consumed with thoughts of tomorrow’s generation. His foundation in Johannesburg is devoted to providing digital skills to school children from poor localities. It teaches them robotics and coding. The 14–15-year-old boys and girls are bubbling with enthusiasm when they visit the foundation. The next step is to empower these kids with AI skills.

 

His friend Mathews Phosa has another approach. He was a freedom fighter like President Motlanthe. In the last two decades, he has become a king maker. He brought down Thabo Mbeki from the presidency of the country, and helped Jacob Zuma and Cyril Ramaphosa to become president. He is a highly successful politician, lawyer, businessman, and author. He is also concerned about South Africa’s place in the global technological race. He wants the ministers of the government to have the right policies that will make South Africa a rule shaper rather than a rule taker when global rules of AI are negotiated.

 

There is a common message in Motlanthe’s bottom-up strategy and Phosa’s top-down strategy. South Africa has many problems of the past that are carried forward to the present. But it should not lose the sight of the future. They realize that Artificial Intelligence is an important driver of this century.

 

The South Africans are angry that the present AI models carry racial bias. Ask Grok, Chat GPT or Gemini to describe a successful businessman. The model will portray the image of a white guy wearing a tie. It will not project Patrice Motsepe or Mathews Phosa. Ask the model to portray the image of a criminal. It is quite likely to present the image of a black man wearing a torn shirt.

 

The South Africans want to fight racial bias in the new technology. Pelonomi Moiloa and Vukosi Morivate have developed a Small Language Model. It is trained on the South African languages. This is a trend that we can observe in many countries including India, Korea, Brazil, and Saudi Arabia. The AI models trained on local language data are becoming popular. But the question of the world’s leading models displaying racial, cultural and gender balance remains as it is.

 

There is no alternative but for these countries to come together to shape the global rules for shaping the future of Artificial Intelligence. The South Africans are acutely aware that they must play a global role, and they cannot do it if they only go on complaining about racial bias. They realize that they must go beyond their own concerns to address the problems created by the advanced AI systems that pose existential risks if the powerful models go out of control, or if they are used to develop weapons.

 

I met Alfred Mashishi in a restaurant in Pretoria over a lunch of mielie pap. He is a high official in the ministry of digital technology. He has authored several policy papers on the future of AI. Some of them are being discussed in the parliament. The message in all these documents is clear: “South Africa wants to be a significant player on the global stage. This involves not only technological advancements but also influencing global AI ethics and governance frameworks.” The ministers, high officials, and politicians I met in Cape Town and Pretoria shared this aspiration.

 

South Africa is uniquely positioned to combine the immediate local concerns with futuristic global threats. The South Africans have a feeling of “Oh, we have been here before.” They won independence from Apartheid only 40 years ago. They do not want Algorithmic Apartheid to rule the world. Also, they once had nuclear weapons and they dismantled them voluntarily. They do not want AI to transform into another dangerous weapon like a nuclear weapon. The South Africans see that the future of AI must be shaped by their experience in their own history.

 

They are also proud of their philosophy known as “Ubuntu”. It means “I am because we are.” If advanced AI systems threaten the existence or even stability of humanity, all people on the planet will be affected. If all are devastated, I am also finished. The critiques argue that South Africans do not follow “Ubuntu” in their everyday life. There might be some truth in it. But South Africa seems to follow “Ubuntu” in its foreign policy.

 

President Cyril Ramaphosa will preside over G-20 summit in November. South Africa is also an active member of BRICS and many pan-African forums. It has opportunities to articulate a new “Ubuntu” approach to AI governance in the world. But it is not going to be easy. As the host of G-20, South Africa has received suggestions from almost 300 expert groups. The recommendations on Artificial Intelligence stop at inclusion and access to technology. They do not recognise that AI is moving in the direction of highly advanced systems, and that it can potentially take control over human society. Just as South Africa showed courage to reject Apartheid and its nuclear weapons, it will have to play a global role to promote the principles of fairness, justice, and responsibility to humanity.