When AI Meets Society

How do we deal with the social implications of AI’s rapid advance?

This week, Timnit Gebru, a prominent researcher in machine learning, the former co-lead of the Ethical AI team at Google, and the co-founder of Black in AI, gave a talk at the University of Chicago’s Center for Data and Computing Distinguished Lecture Series.

Academic researchers listen to a lot of talks, perhaps as many as 5–10 every week in “peak season”. Every now and then, there is a talk that profoundly changes the way we think about our entire approach to research, to our philosophy to the types of problems we choose. Timnit’s talk was one such talk. It is a must-watch for any computer scientist.

This spring, we at the University of Chicago’s Center for Data and Computing have been hosting a speaker series entitled “Bias Correction: Solutions for Socially Responsible Data Science”, which has hosted a number of great speakers that have explored some of the social implications of machine learning, including how to design machine learning systems that take these various social effects into account.

Timnit’s talk highlighted several important lessons that are important for any computer science researcher who is working on…