A survey by Nairobi-based research consultancy Nendo suggests that generative AI is being used by digitally active Kenyans for more than routine work, with respondents turning to chatbots for emotional support, personal guidance and matters they keep private from other people.

Table Of Content

Private and emotional use
Trust and local relevance
Limits of the findings

Nendo surveyed 1,200 digitally literate, AI-aware people across 18 Kenyan counties. Of those, 1,161 completed the detailed module on AI use and attitudes.

The findings indicate that respondents use these tools for a wide range of tasks, including business plans, financial guidance, academic coursework, translation, lesson plans, cover letters and research papers. The survey also examined use in personal circumstances, levels of trust, cultural compatibility and performance in Kiswahili and Sheng.

Private and emotional use

According to the survey, 46.3% of respondents had used AI for matters they kept secret from others. A larger share, 72%, said they had used it for emotional support or to navigate personal circumstances. Among that group, 54.1% reported doing so weekly.

The survey did not classify every personal query as clinical therapy, and it did not assess the safety or factual accuracy of the responses participants received. However, the results point to conversational tools being used for sensitive issues as well as administrative, educational and financial tasks.

Use varied by age and gender. The proportion reporting emotional or personal use was 76% among respondents aged 18 to 24, rising to 81.4% among those aged 25 to 34. It fell to 63.8% among people aged 35 to 44 and was below 50% among respondents aged 45 and older.

Women reported this type of use more often than men, at 76% compared with 68.4%. Nendo’s findings did not establish why the age and gender differences occurred.

Trust and local relevance

The survey also recorded notable levels of confidence in AI for consequential decisions. Some 57.9% of respondents expressed high or complete trust in it for career decisions. The corresponding figures were 55.7% for legal issues, 53.5% for financial choices and 50% for health matters.

Family-related guidance attracted a lower level of equal trust, at 39.9%. The explanation for this difference was not explicitly tested by the survey.

Views on whether AI reflects Kenyan life were mixed. While 72.5% of respondents said AI models broadly understood daily life in Kenya, 25.1% said they had received recommendations completely at odds with local culture or values.

Language performance was another area of concern. Only 34.9% rated AI responses in Kiswahili or Sheng as very good or excellent, while 35.8% considered them fair or poor. The findings suggest that wider use of these tools does not necessarily mean they consistently reflect local language or cultural contexts.

Limits of the findings

Nendo’s sample was younger, better educated and more active online than the average Kenyan adult. The survey is therefore not nationally representative of Kenya, and its results should not be treated as a picture of the entire population.

Even so, the findings highlight how the same interfaces used for drafting documents and finding information are also becoming channels for personal disclosure. Nendo’s conclusion is that meaningful localisation must capture “the difference between Kenya and a stereotype.”