Understanding the Perceptions and Practices of the Machine Learning Professionals in Bangladesh

A growing body of research focuses on the machine learning professionals in the Global North, but the perceptions and practices of the professionals in the Global South have remained understudied. For this work, we interviewed 15 AI/ML professionals from Bangladesh to gain insights into their motivations, perceptions, and practices when working with machine learning tools and technologies. Our findings reveal that the machine learning professionals in this region prefer labeled datasets for their work and have lack of knowledge regarding annotation, design, or representation bias. Their responses also indicate that resource constraints and systematic or organizational barriers limit their capacity to address environmental concerns as well as explainability and privacy issues. Our initial findings highlight the importance of a large-scale study with the machine learning professionals in the Global South.