In a resounding call that has electrified Africa’s tech ecosystem, Professor Folasade Ogunsola, Vice Chancellor of the University of Lagos, declared today: “We cannot and will not outsource the intelligence of our continent.”
Speaking at the university’s inaugural Africa-Native AI Summit, Ogunsola warned that reliance on foreign AI models risks embedding systemic bias, linguistic erasure, and strategic dependency. “These systems speak fluent English, but they are mute in Hausa, Yoruba, or Zulu,” she said. “They optimize for Silicon Valley suburbs, not Lagos traffic or rural clinics in Sokoto.”
Her remarks come as GSMA, in partnership with six major African telecom operators MTN, Airtel, Vodacom, Orange, Safaricom, and Ethio Telecom launched the “AI Language Models in Africa, By Africa, For Africa” initiative in Kigali last week. The $180 million program aims to develop foundation models trained on over 2,000 African languages, using locally sourced data and compute infrastructure.
A Continent-Wide Chorus:
Ogunsola’s stance is not isolated. In Nairobi, iHub and Konza Technopolis have pledged to prioritize open-source African LLMs. In Cape Town, the African Institute for Mathematical Sciences (AIMS) announced a new PhD track in contextual AI, focusing on low-resource languages and energy-constrained environments.
Even policymakers are listening. Nigeria’s Minister of Communications and Digital Economy, Bosun Tijani, tweeted this morning:
“AI sovereignty is national security. We will fund African datasets, African GPUs, and African ethics.”
The Technical Gap and the Fix
Current global models like GPT-4 and Llama 3 perform poorly on African tasks: Speech recognition: <40% accuracy in Swahili vs. 98% in English
Translation: Igbo-English pairs fail 70% of the time
Cultural reasoning: Models misidentify local proverbs, kinship systems, and market dynamics The GSMA initiative counters this with:
Masakhane 2.0 – a pan-African NLP community scaling to 500+ contributors AfriCompute solar-powered edge clusters in Ghana, Kenya, and Rwanda Data trusts community-owned repositories with profit-sharing for contributors.
Ethics with Roots: OMOLUABI-AWARE Goes Live
Complementing the technical push, Black House Media (BHM) officially released the OMOLUABI-AWARE Framework today, the first AI ethics standard built on African philosophical principles. Named after the Yoruba ideal of moral excellence, it mandates:
Cultural audit of training data
Community veto on harmful applications
Transparency in algorithmic decision-making
“BHM will audit any AI deployed in African media,” said founder Ayeni Adekunle. “No more foreign black boxes.”
The Economic Stake:
McKinsey estimates Africa’s AI market will reach $16.5 billion by 2030, but only if local models dominate. “Every dollar spent on imported AI is a job lost in Lagos or Accra,” said Dr. Nneile Nkholise, CEO of iMed Tech and summit panelist.
Next Steps:
November 12: Nigeria launches National AI Dataset Consortium
November 15: South Africa hosts AI Summit with live demos of Hausa and Xhosa chatbots.
November 22: Global Africa Partnerships Forum in Johannesburg to align G20 funding with African AI priorities
As the sun sets over Abuja, one message echoes from campus to capital: Africa will not be a consumer of AI. It will be its architect.
