Discussion about this post

User's avatar
He Sun's avatar

An insightful essay. I very much like the way in which you introduce the studied mathematical problems within a broader social context.

I thought about a related question a while ago. While the AI fairness crucially depends on the fairness of the algorithms, it also relies on the fairness of the training sets. At least, a "fair" training set would make it easier for us/AI to correct an algorithm's bias. I was wondering if one can formally study the fairness of a train set, or by nature its fairness is subject to an individual's judgement. Are you aware of related work on this?

Expand full comment

No posts