An investigation has found that an AI-based system used in Kenya’s new healthcare reform is systematically overcharging poorer citizens while underestimating contributions from wealthier households, raising concerns about fairness and transparency. The system, introduced under President William Ruto’s health reform agenda launched in 2024, was designed to replace the old national insurance model and extend coverage to Kenya’s large informal workforce. However, instead of using advanced generative AI, it relies on a predictive machine learning model based on proxy means testing, which estimates household income using indicators such as housing type, assets, and living conditions. Critics and researchers say the algorithm is flawed and opaque, often misclassifying poverty levels and setting unaffordable premiums for low-income families, sometimes consuming up to 10–20% of their earnings. As a result, many Kenyans are reportedly unable to access healthcare, with some hospitals turning away patients who cannot pay. Experts argue that while the system was intended to improve inclusion and efficiency, its design choices have instead deepened inequality and undermined public trust in the healthcare reforms.
Read more here
