Designing the User Experience of Machine Learning

Abstract

Machine learning systems are found everywhere. They support everyday technology: search systems, social media news feeds, fashion advice apps. But machine learning is also used in extremely high impact prediction and decision making systems to predict credit scores, health diagnoses, and allocate child protective services. Unfortunately as algorithmic decision making and governance systems increase in prevalence, importance, and impact, users have felt resignation and fatalism because of their lack of control over these systems. My work has sought to understand how to design systems so that users feel more agency and control. In this talk, I will discuss that work and how can we design machine learning systems so that users can understand them, engage with them, and feel empowered in an increasingly automated world.