The headline number

One statistic has been doing a lot of work in the debate over AI adoption in Africa. In DataReportal's July 2025 survey, 42.1% of Kenyan internet users aged 16 and over said they had used ChatGPT in the past month S1. On that measure, Kenya recorded the highest share of any market the survey covered, and TechCabal put the finding plainly: by one count, Kenya leads the world in AI use S1.

It is a striking figure. It is also a figure with edges, and the edges are where the argument lives.

What 42.1% actually measures

Three things define the number, and each one narrows it.

First, the base is internet users aged 16 and over, not the population at large S1. Kenya's internet-using population is the group most likely to encounter and pay for digital tools, so this is a reasonable base for a technology survey. But it is not a national adoption rate, and it should not be read as one.

Second, the question is about ChatGPT specifically, not artificial intelligence in general S1. ChatGPT is the best-known consumer chatbot, which makes it a useful proxy. It is not the same as measuring AI use in businesses, in software products, or through other assistants.

Third, the response is self-reported: people said they had used the tool S1. Surveys of behaviour depend on recall and on how respondents interpret the question.

None of this makes the finding wrong. It defines what the finding can be compared against. That is the point TechCabal makes in its framing: who is actually using AI in Africa depends on whom you count S1.

Why the counting question is a money question

For anyone deciding where to spend, the definition changes the size of the prize.

Advertisers, app developers, telcos, employers and training providers all use adoption data to size a market. If roughly two in five Kenyan internet users have tried ChatGPT in a given month, that points to a consumer base that is already familiar with the category. If the same survey question were asked about AI tools broadly, or about paid subscriptions, the picture would look different.

The comparison problem is sharper still. Rankings of countries only hold if the surveys behind them ask the same question of the same group. A survey of internet users aged 16 and over is not interchangeable with one covering all adults, and a question about ChatGPT is not interchangeable with one about AI assistants in general. When those bases differ, a leaderboard can move without anything changing on the ground.

For readers weighing where AI spending will land, the practical lesson is to look past the headline percentage to the methodology line underneath it.

What to watch next

  • Whether later surveys repeat the ChatGPT question with the same age range and base, so the 42.1% can be tracked over time rather than treated as a one-off snapshot S1.
  • Whether future research widens the definition to cover other AI tools, business adoption and paid usage, which would test how much of the Kenyan lead is about ChatGPT and how much is about AI more broadly.
  • Whether outlets reporting such surveys publish the base alongside the percentage. The gap between "share of internet users" and "share of the population" is the most common way adoption numbers get overstated.
  • Whether the finding is corroborated by independent surveys, since the case for Kenya's position currently rests on a single published measure S1.