Together, the projects cover more than 50 African languages, dialects and sign languages spoken by more than 500 million people across 47 countries. The selected languages range from widely spoken languages such as Kiswahili, Hausa, Amharic, Wolof and Luganda to less digitally represented languages including Awngi, Tjwao, Doma, Medumba and Ghomala’. At Makerere University in Uganda, researchers will build what the initiative describes as a maternal and child health speech dataset covering Luganda, Acholi, Dholuo, Dinka and Juba Arabic. The data problem behind Africa’s AI gapOne of the biggest barriers to developing effective AI for African languages is the lack of high-quality digital data. That diversity points to a broader shift in African AI development: the focus is moving from simply adopting AI technologies to building the underlying infrastructure needed to make those technologies relevant to African users.