The future of AI in healthcare in Africa

AI in healthcare is having its moment globally. In Africa, where doctor-to-patient ratios are a fraction of OECD norms, the temptation to 'leapfrog' with AI is enormous. Some of that excitement is justified. Some is dangerous.

The justified part: AI is already credible at narrow, image-heavy tasks. Tuberculosis screening from chest X-rays, diabetic retinopathy detection from fundus photos, early-warning scores from vital signs. These tools can extend the reach of a single radiologist or ophthalmologist tenfold, which matters enormously in countries with a handful of specialists per million people.

The dangerous part: large language models confidently dispensing medical advice to patients with no clinician in the loop. The failure mode is not that the model is usually wrong — it is that it is usually right, which trains everyone to stop checking. In a context where there is no malpractice infrastructure to catch the bad cases, that is a serious risk.

Our position is that AI belongs inside the clinical workflow, behind a licensed clinician, augmenting rather than replacing judgment. Used that way, it can give an African specialist the leverage of a small team. Used carelessly, it can give patients the illusion of care without any of its substance.