Kasia Chmielinski will moderate and participate in a panel discussing ways to eliminate bias in machine learning. They are
project lead of the Data Nutrition Project, a part of the Assembly program at the Berkman Klein Center at Harvard University & MIT Media Lab.

Tackling bias in the datasets used to build AI is key to rooting it out from the very beginning of the AI development process. Based on the USDA nutrition labels to show necessary information about a food as succinctly as possible, the dataset nutrition label shows key ingredients, like meta-data and populations as well as unique features pertinent to the set. Prior to the Data Nutrition Project, they worked at the U.S. Digital Service and the Scratch project at the MIT Media Lab.

The Data Nutrition Project is still in the early stages. Read about their progress in their own words:

With one prototype for people to see, and your paper published May 2018, are you working on more labels? More papers?

“We have been focused on convening cross-industry conversations for the last few months as we continue to work on the label and prototype behind the scenes. We’ve spoken with standards bodies, government agencies, academics and industry about the label. We’re also conducting workshops about AI more broadly to help spread the word about the importance of data nutrition. Actually, the Data Nutrition Team – myself and colleagues Sarah Newman and Matt Taylor – will be conducting a free workshop during SXSW. Folks should come by.”

Have you had the opportunity to build more prototypes?

“We don’t have anything public to share yet but have collected feedback and had several conversations about the initial prototype to help us iterate towards the next version.”

What are some of ways people can help with this project?

“We’re always looking for like-minded folks. If you’re a data scientist, machine learning expert, designer or workshop facilitator, we’d love to talk. Drop us an email at info@datanutrition.org or come find us at SXSW.”

For more information, check out Data Nutrition Project, Demystifying Artificial Intelligence workshop and Bias in, Bias out: Building a better AI

Mar 11, 2019 at 12:30PM – 1:30PM at Fairmont Manchester AB