AI Startups Mapping in Africa

(iv) Description of AI Formal Education and AI Subfields Specialization The study findings indicate that 63% of the respondents have formal education on AI, with 37% at the final stage, 27% at the intermediate stage, and 36% at the initial stage of AI adoption. The AI Start-ups in Africa have focused on the following AI Subfields Specializations; whereby majority apply Machine Learning techniques in their production at 30% ; Deep Learning at 17; Natural Language Processing at 15% ; Computer Vision at 12% ; Generative AI at 10%; Reinforcement Learning at 9% ; and Robotics at 7% as detailed in Figure 7.

Yes|No

37 % 27 %

7 % Robotics

Final Stage

63%

37%

Formal Education on AI

9 % Reinforcement learning

Intermediate Stage

36 %

Final Stage

30 % Machine Learning

10 % Generative AI

AI Adoption

Description of AI Subfields Specialization in Africa

12 % Computer Vision

The AI Start-ups in Africa have focused on the following AI Subfields specializations; whereby majority apply Machine Learning techniques in their production at 30%; Deep Learning at 17%; Natural Language Processing at 15%; Comput- er Vision at 12%; Generative AI at 10%; Rein- forcement Learning at 9%; and Robotics at 7%

15 % Natural Language Processing

17 % Deep Learning

Figure 7: Description of AI Formal Education and AI Subfields Specialization.

ics processing units (GPUs), tensor processing units (TPUs), and others, increases with growing AI algorithm complexity. For the past decade, the world has witnessed unparalleled levels of in- vestment in digital infrastructure which has been brought due to increase in smartphone penetra- tion, industry digitalization and the rise of cloud computing, however the rapid evolution of AI has

(v) Description of Needed AI Infrastructure The emergence of AI is fueling the surge in the need for digital infrastructure and comput- ing power worldwide. The degree of computing power is crucial in developing and deploying AI models, particularly those involving computations and massive datasets. The need for high-perfor- mance computing (HPC) devices, such as graph-

20

Made with FlippingBook - Online Brochure Maker