Why AI needs a "nutrition label" | Kasia Chmielinski

This podcast episode explores the challenges and risks associated with algorithmic systems, using the analogy of sandwiches to illustrate the complexity of understanding and addressing issues in artificial intelligence (AI). The speaker discusses the negative repercussions of AI, the role of data in fueling AI, and the need for understanding data quality. They introduce the Data Nutrition Project, which aims to provide nutrition labels for datasets to enable users to assess data suitability. The section also highlights the challenges of data labeling, lack of AI regulation, and transparency in data sets. The speaker emphasizes the potential risks and harms of generative AI and proposes three principles for corporate engagement with data to promote AI accountability. The ultimate goal is to establish an integrated, algorithmic internet that prioritizes user well-being.