L.030.35470 Algorithmic Discrimination and algorithmic fairness

Lehrende: Prof. Dr. Tobias Matzner

Veranstaltungsart: Seminar

Orga-Einheit: Medienwissenschaft

Anzeige im Stundenplan:

Semesterwochenstunden: 2

Unterrichtssprache: Englisch

Min. | Max. Teilnehmerzahl: - | 40

Kommentar:
Almost every week we hear of a new case of algorithmic discrimination: Algorithms misclassify people as gorillas, they do not recognize dark-skinned faces, they would hire biased towards men etc. However, what does it mean that an algorithm is discriminatory? We usually think of discrimination in terms of a few protected categories like looks, gender, origin, religion, sexuality, political views. Algorithms often use features like Facebook likes, shopping history, websites visited, social network graphs, etc. in almost endless combinations? Can we speak of discrimination in terms of these categories? Do we need a new concept of discrimination? These questions will be discussed in the first part of the seminar. The second part is dedicated to a rather recent research area: Algorithmic fairness. Several attempts have been made to program algorithms in a fair manner. So how can we translate our legal and social ideas of justice and equality into an algorithm? Is that possible at all?

The class will discuss recent development in information technology but does not presuppose any prior knowledge. The focus will be on the medial and social aspects of these technologies.

The course is taught in English.