Skip to main content

Artificial Intelligence and Medicine: African Philosophical Perspectives


This project investigates medical, epidemiological, and public health uses of AI from African philosophical perspectives, using African moral theories and other relevant African philosophical tools to achieve these primary objectives:

  • Providing conceptual grounding for the development of medical artificial intelligence for an African context, which is commonly either neglected or treated as a mere recipient of technology, knowledge, and medical expertise sourced elsewhere;
  • Identifying ways that African philosophical resources can influence global policies on the development and use of medical AI, especially to mitigate ethical issues such as algorithmic bias and other forms of technological malfeasance; and
  • Providing conceptual and methodological insights that are not visible from the conceptual frameworks in which such concepts and methods have usually been developed, and which might never be accessible from a position of historical dominance.

The use of AI within the medical space is not as new as it is might seem. Early uses occurred in the 1970s, when early forms of AI were used to diagnose and treat pathologies such as glaucoma and other infectious diseases by implementing Bayesian approaches. However, with the growing application of machine learning techniques in the second decade of the present century, AI already has come to play a much more significant role in medicine and, increasingly, in public health. Image recognition, in particular, has become extremely effective, and clinicians increasingly rely on machine learning technologies for clinical diagnosis and prognosis of medical conditions.

These remarks, however, apply primarily to Europe, the US, China, and other technologically developed nations. Countries in the Global South, and especially in Africa, are yet to leverage the potential applications of medical AI. Sub-Saharan Africa is the poorest region of the globe, and many of its countries are currently disadvantaged economically and infrastructure-wise. Yet precisely these circumstances create significant opportunities for the deployment of medical AI, which has already been deployed in several sub-Saharan African countries to assist medical practitioners. For example, recent developments such as Ubenwa, a machine learning technology developed in Montreal and used for the detection of birth asphyxia in Nigeria, and the Delft Institute Computer Aided-Detection for Tuberculosis (CAD4TB) used in detecting retinopathy in Zambia and pulmonary tuberculosis in Tanzania and Zambia, have shown how such technologies can greatly extend the reach of limited medical personnel and equipment resources. At the climax of COVID-19, “robot doctors” donated by the United Nations Development Programme were deployed to assist frontline workers in South Africa and Rwanda.

However, while these technologies come with enormous promises in Africa, there are still challenges when it comes to successfully applying medical AI within the African context. These problems all connect with a lack of regional agency in the development, training, and deployment of these technologies. There is a lack of large clinical datasets that can be used in training AI models. The African health sector faces challenges such as poor digitalisation of medical records; as a result, it is difficult to build AI systems using African data. Current AI models deployed to Africa, especially sub-Saharan Africa, come from the global North. Such models imply that the datasets used in developing them are from people with different physiological qualities, which may lead to algorithmic bias in these technologies.

These are practical problems but there are underlying conceptual, methodological, and ethical issues that require philosophical skills to resolve. What is algorithmic bias and how can it be avoided? How should the values and priorities that inform any modelling efforts be set, and how does this problem relate to larger questions about indigenous knowledge systems, paternalism, and liberal universalism? How can colonial power dynamics be better understood and therefore spotted, even—or especially—when they threaten to distort otherwise well-intentioned efforts? Problems like these can only be adequately addressed from African perspectives, and doing so is to lay the foundation for African agency in the development and use of AI for medicine and public health.

Get in Touch

Enquiries about the CPEMH should be sent via the email below.