Understanding Gender Biases in Building Artificial Intelligence Models in the African Context

Country : Uganda

Project’s summary

Machine learning technologies like Automatic Speech Recognition systems, language models, and machine translation systems are developed to work in environments that have gender and other forms of biases and discrimination present.

These biases are present in the datasets that are used for model training and the choices around the ML models. The developed ML models end up reinforcing biases that are embedded in training data.

This research will focus on a data-driven research approach to carry out a scoping study on a deeper understanding of gender biases and inclusivity of AI tools on the African continent. The results will be used to build a framework and guidelines for the mitigation of gender biases in ASR systems through equity in data collection and model creation.

Finally, this work will develop a model governance framework to address gender and ethical aspects around the data and algorithms from which ASR systems are built.

Project implementation zone : Uganda

Project leader’s name Dr Joyce Nakatumba-Nabende

Joyce Nakatumba-Nabende is a senior lecturer in the Department of Computer Science. She is the head of the Makerere Artificial Intelligence Lab, where she leads and has worked on research projects that aim at developing and applying Artificial Intelligence and machine learning methods and tools to improve the quality of life, especially in the developing world. She is a board member of Data Science Africa and also a member of the Educational Advisory Committee of the ACM

Team member’s biographies 

Dr. Andrew Katumba

Andrew Katumba is a lecturer in the Department of Electrical and Computer Engineering in Makerere University. His current research focuses on applying machine learning and data science techniques in areas relevant to the African Context. Some key examples include : 1) the development of a machine learning-aided smartphone platform for cervical cancer diagnosis, 2) the development of a portable machine learning-aided lung ultrasound system, 3) the development of a diagnostic platform for plant diseases 4) building of machine learning-aided NLP tools for low- resourced languages.

Dr. Peter Nabende

Peter Nabende is a senior lecturer in the Department of Information Systems, School of Computing and Informatics Technology, Makerere University. His research interests include Natural language Processing, Artificial intelligence applications for the developing world, and data mining. He has worked on research projects to develop resources and models for applications in low-resourced African contexts. He has also developed a short tutorial on eliminating or mitigating biases in AI training data.

Eric Peter Wairagala

Eric Peter Wairagala is a Research Software Engineer at the Makerere University AI and Data Science Research Lab. He is a highly motivated and passionate individual pursuing his Masters in Computer Science at Makerere University with a focus on AI and Data Science. His research interests cut across the fields of Computer Vision, Machine Learning and Natural Language Processing, with a particular focus on responsible AI. He is a collaborative member of the Masakhane, Data Science Africa, and Deep Learning Indaba communities, which aim to bridge the gap between AI and machine learning in Africa.

Tobius Saul Bateesa

Tobius Saul Bateesa is a Research Assistant and a Software Engineer at Makerere Artificial Intelligence and Data Science Lab. At the Makerere AIR Lab he has worked on projects that use Data science and Artificial intelligence to solve challenges in a developing countries in fields such as agriculture, health, urban planning, ethical and culture preservation. He is currently pursuing his Master of Science in Data Science at the University of East London, where he continues to deepen his expertise.

Carol Kantono

Caroline Kantono is a Research assistant at Makerere University Artificial Intelligence and Data Science Lab. She is pursuing a career in data science having completed her Master’s degree in Business Intelligence and Analytics from University of Applied Sciences Neu-Ulm. Her research interests are in the field of Natural Language Processing (NLP) and Artificial Intelligence, where she completed her research project on sentiment analysis.

Institution’s description

Makerere Artificial Intelligence (AI) lab is a research lab in Makerere University that specializes in the application of artificial intelligence and data science to problems in the developing world. The AI lab has over a decade experience in undertaking novel AI research with a focus on the application to local problems. Driven by our mission “To advance Artificial Intelligence research to solve real-world challenges”, and through strong partnerships with the national organizations and university departments like the Institute of Languages, we have competencies in (a) the development of tools for the collection and curation of datasets for downstream NLP tasks, (b) building computational models for downstream Natural Language Processing tasks, (c) building computational models for image crowdsourcing and real-time surveillance, (b) application of computer vision for crop pests and disease diagnosis.

Link of institution : https://air.ug/