The Ghana Centre for Democratic Development (CDD-Ghana) has unveiled its study report evaluating Gender bias in Artificial Intelligence (AI) applications using household survey data.
AI, known for simulating human cognitive processes, is transforming sectors like healthcare, finance, and education. The study, conducted in collaboration with AidData, assessed gender biases in Wealth estimation AI.
Funded by the United States Aid for International Development (USAID) Equitable AI Challenge, the project aimed to gauge gender bias in AI's wealth measurement using household survey data.
The study utilized data from the Ghana Demographic and Health Survey along with satellite imagery to construct its analysis.
The findings aim to enhance poverty estimates by incorporating gender considerations in resource allocation, impact evaluations, and policy-making. Dr. Edem Selormey, Director of Research at CDD-Ghana, emphasized the growing importance of understanding AI's societal implications, particularly it's potential to exacerbate gender disparities.
Dr. Selormey highlighted that over the past year, AidData and CDD-Ghana delved into gender bias intricacies in wealth estimates generated by AI, utilizing geospatial and DHS data.
The study uncovered biases woven into AI algorithms, demonstrating how models trained on male household data consistently outperformed female-based models.
Dr. Rachel Sayers, Research Scientist at AidData, shared insights on how gender-specific wealth indexes compared to the DHS household wealth index. The study revealed that the DHS wealth index overestimated the wealth of the poorest female households while underestimating the poorest male households.
Dr. Rita Udor, Gender Inclusive Officer at the Kwame Nkrumah University of Science and Technology's Responsible Artificial Intelligence Lab (RAIL), advocated for women's inclusion on AI developers' boards to enhance decision-making quality and ensure equitable technology benefits.
Madam Deborah Dormah Kanubala, Machine Learning (ML) Researcher at Saarland University, Germany, emphasized AI's critical role in medicine, particularly in swift breast cancer diagnosis.
