Can AI Be Harmful? A Conversation with MIT’s Dr. Marzyeh Ghassemi
NEJM AI Grand Rounds
00:00
The Importance of Data in Predicting Bias
A new paper shows that using group attributes in prediction will improve your overall performance for the entire group. But there are ways that you can try to constrain learning such that no subgroup is adversely affected by inclusion of group attributes or demographic attributes. "It's a very critical distinction between monitoring and auditing bias across different race groups," she says.
Transcript
Play full episode
