The Case for Globalizing Fairness: A Mixed Methods Study on Colonialism, AI, and Health in Africa

TL;DR

A scoping review is conducted to propose axes of disparities for fairness consideration in the African context and delineate where they may come into play in different ML-enabled medical modalities and provides practical recommendations for developing fairness-aware ML solutions for health in Africa.

Abstract

With growing application of machine learning (ML) technologies in healthcare, there have been calls for developing techniques to understand and mitigate biases these systems may exhibit. Fair-ness considerations in the development of ML-based solutions for health have particular implications for Africa, which already faces inequitable power imbalances between the Global North and South.This paper seeks to explore fairness for global health, with Africa as a case study. We conduct a scoping review to propose axes of disparities for fairness consideration in the African context and delineate where they may come into play in different ML-enabled medical modalities. We then conduct qualitative research studies with 672 general population study participants and 28 experts inML, health, and policy focused on Africa to obtain corroborative evidence on the proposed axes of disparities. Our analysis focuses on colonialism as the attribute of interest and examines the interplay between artificial intelligence (AI), health, and colonialism. Among the pre-identified attributes, we found that colonial history, country of origin, and national income level were specific axes of disparities that participants believed would cause an AI system to be biased.However, there was also divergence of opinion between experts and general population participants. Whereas experts generally expressed a shared view about the relevance of colonial history for the development and implementation of AI technologies in Africa, the majority of the general population participants surveyed did not think there was a direct link between AI and colonialism. Based on these findings, we provide practical recommendations for developing fairness-aware ML solutions for health in Africa.

Figures (9)

• Figure 1: Categorization of African-contextualized axes of disparities with the type of biases they can induce along the machine learning pipeline. ML pipeline and biases figure modified from chenIY21 and suresh21.
• Figure 2: Perceptions of biases for machine learning tools and their associations with African-contextualized by axes of disparities. a) Experts' (n=57) responses on how likely attributes are to influence AI performance for Africans, b) General population responses (n=672) on whether attributes would cause AI to perform differently (better or worse) or the same for them as for others. They could also indicate whether the attribute was irrelevant. c) Breakdown and ranking of general population participant responses on which attributes would perform worse or better for them. *Better is not necessarily good, it reflects that our participant population is better off than other people in the same country. e.g., Most of our survey participants are skewed towards high literacy and education levels compared to the general population, and so AI may perform better for them than for others.
• Figure 3: Responses by experts during IDIs ($n=28$) and general population participants surveys ($n=672$) on whether they see a connection between AI and colonialism. 57% of experts found a definite link compared to only 9% of general population participants.
• Figure 4: African-contextualized barriers to ML for health by health modality. *=applies to all
• Figure A.1: Example of the questionnaire in the survey