SMEs in Kenya and Nigeria: With the general AI, writing is not the hardest, it is making the result presentable to the customer

A study of 27 employees from 21 SMEs shows that access to AI is not sufficient: verification, traceability and repair of local contraindications are a condition for the announced gains.

Generative AI is often presented as a tool for democratizing skilled work, since it lowers the barrier to the production of professional documents. A team including Najeeb Gambo Abdulhamid and Jacki O'Neill interviewed 27 people working in 21 small and medium-sized enterprises in Kenya and Nigeria between July and October 2023. Observation: having the tool is not enough to take advantage of it.

A "work of circulation"

Participants use AI as their first draft workforce, guide it through iterations, compare several tools, and then devote most of their effort to making drafts defensible before sending them: verification checks, reconstitution of sources when they are missing, and rules for what should never be delegated. The authors call this last link the "release work".

In addition, there is the repair of what they call the context gap : local language, institutional knowledge, credible representation of African people and places, all of which are points on which models regularly fail. The reported productivity gains therefore depend on this invisible work, hence four design avenues for professional tools. The authors specify that the two countries are studied as situated cases, not as a representative sample of SMEs in the Global South.