2022 Ars Electronica Award for Digital Humanity
Artificial intelligence models that are used to make decisions often lead to harmful consequences for communities that are already underrepresented or marginalized. One cause of harm is the data that is used to train the models: biased datasets will replicate the very issues found in the training data. The Data Nutrition Project is an initiative to enable the quick evaluation and interrogation of datasets through the Dataset Nutrition Label. Like a nutrition label for food, the Dataset Nutrition Label conveys information about a dataset and can help mitigate harm caused by problematic data. The team also works on educational initiatives, including a children’s book and podcast. These initiatives are intended to drive awareness about algorithmic risks, as well as current interventions to mitigate them.
Credits
Project lead: Kasia Chmielinski
Research lead: Sarah Newman
Tech lead: Matt Taylor
Engineer: Kemi Thomas, HG King
Data science advisor: Chris Kranzinger
Designer: Carine Teyrouz
Children’s book illustrator: Michael Sherman
Board of Directors: Jessica Fjeld, Mary Gray, Josh Joseph, and James Mickens
With previous support from: The Harvard Data Science Initiative at Harvard University, Digital Lab at Consumer Reports, The Assembly Fellowship at the Berkman Klein Center for Internet & Society at Harvard University, and The Miami Foundation
