Machine Fairness: Algorithms and the Theory of Justice

Catalog description

Artificial intelligence is causing unprecedented changes and disruptions in many sectors of society, raising fundamental ethical and philosophical questions. Although many researchers are currently studying how tasks can be automated efficiently, only a few have discussed how tasks can be automated fairly, to produce the best possible outcome for society. Machine Fairness is an emerging area of research at the intersection of philosophy, machine learning, computer science, statistics, and psychology. This course is designed as a non-technical introduction to the basic problems involved in answering questions about machine fairness. It focuses on a number of applications, including criminal sentencing, predictive policing, self-driving vehicles, autonomous weapons, and healthcare. Although these areas are quite different, they involve similar questions. What biases might enter into algorithms, and what is their origin? How should we evaluate whether the outcomes of an algorithmic process are just? When things go wrong, who should be held accountable? The course addresses these questions by using methods from philosophical ethics.