On November 5, Ben Chapman, Executive Director of the Legal Analytics & Innovation Initiative, offered a workshop on algorithmic bias. "Algorithmic bias" occurs when errors in a computer system create unfair outcomes. There are many sources of algorithmic bias, ranging from the way that data is collected or selected to the way that algorithms are chosen and implemented.
In the workshop, we discussed a number of sources of algorithmic bias, drawing on materials from Berkeley's Daylight Lab at the Center for Long-term Cybersecurity. Attendees at the workshop then participated in a lab that explored bias in a heavily-used healthcare risk scoring algorithm.
The workshop recording may be viewed at this link.
The Legal Analytics & Innovation Initiative plans to offer another workshop on addressing bias in machine-learning models in early spring 2022.