On 6 September 2024, during the week‑long Deep Learning Indaba in Dakar, Senegal (1–7 Sept), Lanfrica co‑organised the full‑day workshop titled “Empowering African Voices in AI: Data, Models & Innovation”. The session took place at Amadou Mahtar Mbow University and formed part of the official DLI workshop track.

Why this workshop mattered

Although AI is transforming sectors worldwide, its benefits remain unevenly distributed; under‑resourced communities often lack the data, legal clarity, and tooling needed to build language‑inclusive technologies. The workshop set out to close that gap by focusing on ethical, culturally aware data practices, fair licensing, and community‑driven model sharing, placing African voices at the centre of AI development.

Objectives at a glance

Participants worked toward five concrete goals:

1. Master efficient, ethical data collection techniques.
2. Demystify proprietary, open, and open‑source licenses for datasets and models.
3. Deepen understanding of AI policy and data rights in Africa.
4. Co‑create multilingual, context‑rich datasets during hands‑on sessions.
5. Explore sustainable data‑economy models, including synthetic data generation.

Voices in the room

Invited experts brought perspectives from research, policy, and industry, including Ndapewa (Wilhelmina) Nekoto (Masakhane), Dr Chinasa T. Okolo (Brookings Institution), Leonida Mutuku (LDRI), Dr Lilian Wanzare (Maseno University), and Chloe Burke (Meta). Keynotes and talks also featured Balkissa Ide Siddo (Meta) and Aisha Iqbal (Meta).

Highlights from the agenda

• Opening remarks by Chris Emezue framed the event around data empowerment.
• Keynote: Harnessing Open‑Source Innovation in NLP set the tone for collaborative solutions.
• Data licensing tutorial led by Dr Chijioke Okorie unpacked practical routes to share data responsibly. Finally they unveiled the very first African dataset license — the Nwulite Obodo Open Data License.
• Panel on Ethical AI balanced innovation with data rights, featuring Vukosi Marivate alongside invited speakers.
• Immersive activity guided by Wilhelmina Nekoto produced draft blueprints for culturally relevant corpora.
• Behind‑the‑research lightning talks gave emerging scholars five minutes to spotlight work and receive feedback.

Key takeaways & outcomes

• Community‑first data design. The activity facilitated by Wilhelmina reinforced that context and consent trump sheer volume.
• Actionable ethics. The panel crystallised guidelines for balancing commercial innovation with community data rights.
• Recognition. Every full‑day participant received an official certificate, a first for many early‑career researchers.

Keep building with us

If you’re passionate about ethical, inclusive AI for African languages, whether as a linguist, developer, or policymaker, Lanfrica invites you to join the conversation. Together we can ensure every African voice is not only heard but actively shaping the continent’s digital future.