The Combatting Bias Toolkit is designed to help AI/ML practitioners and managers identify, understand, and mitigate sources of bias in their data, algorithms, and processes. The toolkit is open source and available on GitHub to disseminate tools and resources as broadly as possible and to encourage teams across the federal government to implement bias mitigation in their projects.
Currently in beta, the Bias Toolkit is comprised of several tools that seek to address the federal government’s AI mission through performance-driven, accurate, and transparent practices. The toolkit is growing, and the project team is seeking opportunities for growth through new partnerships. Teams across the federal government are encouraged to contribute problem statements and ideas.
The toolkit was originally funded by the 10x Program at the General Services Administration and developed by xD at the U.S. Census Bureau to promote best practices and the usage of practical, open-source tools to meaningfully address bias in government data and algorithms.
To contribute resources or case studies, learn more, or for more information about this project or xD (www.xd.gov), please reach out to email firstname.lastname@example.org (mailto:email@example.com).