Author: N/A
Publisher: The Data Nutrition Project
Publication Year: N/A
Summary: The following organization’s mission is to empower “data scientists and policymakers with practical tools to improve [artificial intelligence (AI)] outcomes.” This page outlines the problem with AI algorithms as garbage in, garbage out. They argue that training datasets need to be assessed based on standard quality measures that are both qualitative and quantitative. This organization has developed a tool called the Dataset Nutrition Label that creates a standard label for interrogating datasets. This Label can help users determine whether the dataset is a good fit or “healthy” for a particular statistical use case. This is to mitigate harms caused by statistical systems, providing information about the dataset mapped to common use cases. The organization is comprised of a group of researchers and technologists working to tackle the challenges of ethics and governance of artificial intelligence as a part of the Assembly program at the Berkman Klein Center at Harvard University and & MIT Media Lab.