How to design algorithms with fairness in mind
The Future of Everything
00:00
Is There a Difference Between Fairness and Individuality in Machine Learning?
In addition to my profit, I have duties to serve other societal priorities. And then you can actually build that in mathematically to the function that's deciding whether a certain machine learning algorithm is performing well or not. What we said is that fairness towards group is very, very weak. You can abuse each one of those notions. It's easy to abuse it. We try to give some guarantees to individuals. That works specifically about how to treat similar people similarly,. But we gave a mathematical definition that makes it possible for algorithms to enforce it on algorithms.
Transcript
Play full episode
