Kenya's AI healthcare system overcharges poorest citizens, investigation finds
An investigation found that Kenya's machine learning algorithm for calculating healthcare contributions, launched in October 2024, systematically overcharged the poorest citizens while undercharging the wealthiest. The flawed means-testing system has sparked protests, reportedly raising costs for millions in the informal economy who were promised affordable access to care.
What happened
A predictive machine learning algorithm used by Kenya's healthcare system launched in October 2024 has systematically overcharged the poorest Kenyans and undercharged the wealthiest, an investigation found. The system determines healthcare contributions through a means-testing process that collects household information from citizens.
The algorithm overestimated incomes of poor Kenyans while underestimating incomes of wealthy citizens, according to an investigation by Africa Uncensored, Lighthouse Reports and The Guardian. The healthcare system was intended to replace Kenya's decades-old national insurance system and expand access to care for the informal economy.
Reported impact
The system has sparked protests and anger as healthcare contributions for millions of people are calculated via a formula described as flawed with almost no transparency. Citizens report being confused and fearful during the registration process, with some fearing they are under investigation.
Index analysis
The algorithm's bias against the poor and toward the wealthy suggests the predictive model failed to account for income measurement challenges in Kenya's large informal economy, which comprises 83% of the workforce including day labourers, hawkers, farmers and non-salaried workers.
Governance controls
The means-testing process collects household information including details about toilet type, roof material and radio ownership through digital questionnaires. However, sources reported the system has almost no transparency.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.