Interpretability, Fairness, Controllability - and the process of getting to Responsible AI
Machine learning holds the promise of transforming our lives for the better, but these systems need to be developed responsibly. There have been many calls that stakeholders and users of these systems need to understand how these systems work so that they are trusted appropriately and used effectively. Explainable AI (XAI) has made great strides towards making these systems more interpretable but so far involvement of stakeholders in fairness and control has lagged behind. In this talk I will cover some of my own work in this area, and the challenges to be overcome to build responsible AI.
