In this talk we will discuss some recent concepts of fairness that arise
in the application of artificial intelligence methods at scale. These
include, adaptive policing, judicial decisions, loan applications,
university admissions, advertising and fake news. The two main concepts
of fairness will be introduced at a high level and finally we will
conclude with a description of our group's work on a Bayesian analogue
of those concepts, which explicitly takes into account model
uncertainty. This allows it to take into account informational feedback,
i.e. additional discrimination introduced through biased sampling.
Speaker: Christos Dimitrakakis
Meeting place: Ole-Johan Dahls hus (map), SIRIUS (8459)
Meeting time: November 22nd, 2018; 10:15-12:00