The elusiveness of ethics: encoding fairness in an unfair world

While the problems of data bias and algorithmic bias have hit the mainstream, the mathematical approaches to tackle this problem are mostly restricted to a rarified academic discourse. In this talk you’ll get a birds-eye view of the state of ML Fairness, from checklists and frameworks to developer tools and mathematical definitions of bias and fairness. We’ll wrestle with the elephant in the room: How do we encode fairness into our models when we can’t precisely define our ethics as a society, a team, or sometimes even as individuals?