An ambitious attempt by William Ruto’s government to overhaul Kenya’s healthcare system using artificial intelligence has come under sharp criticism after an investigation found that the reform is disproportionately burdening the country’s poorest citizens, instead of improving access to affordable care.
The new system, rolled out in October 2024, was designed to replace Kenya’s long-standing national health insurance scheme and expand coverage to millions working in the informal sector, including farmers, labourers and small traders who make up a significant share of the workforce. It was presented as a key pillar of the government’s push for digital transformation, with assurances that no citizen would be excluded from healthcare services.
However, findings from a joint investigation by Africa Uncensored, Lighthouse Reports and The Guardian indicate that the technology underpinning the reform has produced the opposite effect. The system relies on a predictive algorithm that estimates household income through proxy indicators such as housing conditions, ownership of assets, and family size. These inputs are gathered through extensive household surveys conducted by field workers, who ask residents a range of personal and sometimes intrusive questions.
According to the investigation, the algorithm has consistently misjudged people’s financial capacity. In many cases, low-income households have been classified as better off than they are, resulting in significantly higher healthcare contributions. At the same time, wealthier individuals have often been assessed as poorer, allowing them to pay less than their fair share.
For many families already struggling to meet basic needs, the revised contributions have proven unaffordable. Reports indicate that some households are being asked to pay between 10% and 20% of their income toward health coverage, forcing them to make difficult choices between medical care, food and housing. As a result, a growing number of people are either delaying treatment or avoiding hospitals altogether due to inability to pay.
The consequences have been severe. Community workers involved in registering households have reported cases of patients being denied treatment or remaining at home without care because they could not meet the required payments. Public frustration has also intensified, with widespread complaints on social media and mounting criticism from experts and civil society groups.
Health economists involved in advising the government have acknowledged inherent flaws in the system. Some have indicated that policymakers were aware of limitations in the model but proceeded regardless, prioritising accurate assessment of wealthier populations even if it meant overestimating the incomes of poorer households.
The approach itself is based on proxy means testing, a method widely used in development programmes to estimate income through indirect indicators rather than direct financial data. While commonly applied in various countries, critics argue that it is prone to errors and can entrench inequality when used as the basis for essential services such as healthcare.
Despite earlier warnings in internal assessments that the system could be inequitable and outdated, it was implemented nationwide. Since then, the reform has faced growing resistance, with many citizens refusing to pay and hospitals reporting financial strain due to reduced contributions and unpaid bills.
The controversy highlights broader concerns about the use of artificial intelligence in public welfare systems, particularly in low- and middle-income countries where data limitations and governance gaps can amplify risks. While digital tools are often promoted as a means to improve efficiency and expand access, the Kenyan case underscores the importance of transparency, accountability and robust safeguards to ensure that such technologies do not deepen existing inequalities.
