Flaws in Kenya’s AI-driven health reforms driving up costs for the poorest

2026-05-04 · Source: AI (artificial intelligence) | The Guardian · Field: Health & Wellbeing — Healthcare Systems & Policy, Public Health & Epidemiology, Medical Devices & Health Technology · Depth: Novice, medium

Summary

Kenya's new "AI-powered" healthcare system, launched in October 2024 by President William Ruto, uses a predictive machine learning algorithm to determine healthcare contributions. Intended to expand access for the 83% of the workforce in the informal economy, an investigation by Africa Uncensored, Lighthouse Reports, and the Guardian found the system systematically overcharges the poorest Kenyans while undercharging the wealthiest. The algorithm, based on proxy means testing (PMT), estimates income using household possessions and circumstances, leading to many low-income individuals being charged 10% to 20% of their meager earnings. This has resulted in widespread protests, with critically ill people unable to afford treatment. A pre-implementation report from IDinsight had already identified the system as "inequitable, particularly for low-income households," yet it was deployed, leading to only 5 million of 20 million registered people regularly paying premiums.

Key takeaway

For CTOs and VPs of Engineering evaluating AI/ML solutions for public services, this case highlights the critical need for rigorous, independent auditing of algorithmic fairness and accuracy, especially concerning vulnerable populations. Your teams must ensure that predictive models, particularly those using proxy means testing, are validated against diverse socioeconomic data and that their decision-making processes are transparent to prevent systemic inequities and maintain public trust. Deploying flawed systems, even with good intentions, can lead to significant social unrest and operational failure.

Key insights

An AI-driven healthcare system in Kenya disproportionately burdens the poor due to flawed proxy means testing.

Principles

• Proxy means testing often misclassifies income levels.
• Opaque algorithms erode public trust in government services.

Method

The system employs proxy means testing (PMT), collecting data on household possessions (e.g., toilet type, roof material, radio ownership) via digital questionnaires to estimate income and calculate healthcare premiums.

In practice

• Audit algorithmic systems for equitable outcomes.
• Prioritize transparency in public service algorithms.

Topics

• Kenya Healthcare Reform
• AI-driven Healthcare
• Proxy Means Testing
• Healthcare Affordability
• Social Health Authority

Best for: CTO, VP of Engineering/Data, Director of AI/ML, Policy Maker, AI Ethicist, Consultant

Related on AIssential

• Dozens of AI disease-prediction models were trained on dubious data — AI models for disease prediction are being trained on potentially fabricated public datasets, risking patient harm.
• Expanding our AI and Healthcare Portfolio — AI deployment in healthcare, driven by profit, poses significant risks to patient safety, worker dignity, and democratic oversight.
• The Download: the hantavirus outbreak and Musk v. Altman week 2 — AI advancements present both significant societal risks, like mass surveillance, and potential benefits, such as military robotics.
• From Radiology to Drug Discovery, Survey Reveals AI Is Delivering Clear Return on Investment in Healthcare — Healthcare and life sciences are rapidly moving from AI experimentation to execution, driven by strong ROI and open-source adoption.
• Fault Lines: Navigating Ethics and Responsible AI Where National Policy Meets Local Practice in Public Sector Transformation — Implementing responsible AI in UK public services faces critical national-local policy gaps and five interconnected challenges.
• The Challenges of Balancing AI Compliance and Technological Innovations in Critical Sectors: A Systematic Literature Review — Balancing AI innovation with compliance in critical sectors faces fragmented regulations, SME burdens, and misaligned governance, requiring strategic solutions.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI (artificial intelligence) | The Guardian.