South African President Cyril Ramaphosa warned that artificial intelligence could deepen existing inequalities if development fails to reflect African communities, languages, and realities. Speaking at the South Africa and the AI Opportunity Conference in San Jose, California, on Wednesday, Ramaphosa called for the development of local datasets and research.
Underrepresentation in AI systems
Ramaphosa stated that African languages, histories, cultures, and experiences remain significantly underrepresented in the data used to train leading AI systems. He cautioned that models lacking comprehension of languages such as isiZulu, isiXhosa, Sesotho, Sepedi, Setswana, Tshivenda, or Xitsonga will not serve all people equally.
The President warned that systems developed without an understanding of African farming conditions, disease patterns, educational challenges, or public administration may not provide the solutions communities need. He said South Africa has an opportunity to strengthen research in African languages, build local datasets, and adapt AI technologies responsibly while protecting personal privacy.
Cross-sector opportunities
Ramaphosa highlighted that AI opportunities extend across multiple industries beyond technology companies. In agriculture, he noted the technology could help farmers manage water efficiently, identify crop diseases, and improve yields, while in healthcare it could support earlier diagnosis and improve medical information management.
In mining, Ramaphosa cited potential benefits for geological exploration, mineral recovery, and equipment maintenance. He also challenged stakeholders to consider using AI to assist township and rural schools, help municipalities detect water losses, and improve public service delivery.
Workforce risks and constitutional values
While noting the technology's promise, Ramaphosa cautioned that technological progress does not automatically guarantee inclusive development. For a country confronting high unemployment, poverty, and inequality, he warned that automating tasks could disrupt existing industries and displace workers.
To mitigate those risks, Ramaphosa called for expanded training in computational thinking, mathematics, and science, alongside retraining programmes for workers affected by automation. He emphasized that national AI development must align with the values of South Africa's Constitution, protecting human dignity and citizen rights against threats including discrimination, disinformation, and fraud.
Key facts and where they come from
- Ramaphosa cautioned that AI could deepen existing inequalities if development fails to reflect African realities and languages.
artificial intelligence (AI) could deepen existing inequalities if its development fails to reflect the languages, experiences and realities of African communities.
- African languages, cultures, and histories remain significantly underrepresented in data used to train leading AI systems.
African languages, histories, cultures and experiences remain significantly underrepresented in the data used to develop leading AI systems.
- AI systems developed without understanding local conditions may fail to provide needed solutions.
A system that has not been developed with an understanding of African farming conditions, disease patterns, educational challenges or public administration may not provide the solutions we need.
- South Africa faces risks of workforce disruption and job displacement from AI automation.
For a country already confronting high unemployment, poverty and inequality, these risks cannot be ignored
The open question: How can AI developers effectively gather representative African language datasets without compromising data privacy?
Tell us what you think · Discuss it in the AI room
Reader poll: Can AI systems succeed in emerging markets without local language training?
A show of hands from readers, not a scientific poll. Say why in the comments.
