Stories about Algorithm Ethics

Interactive web visualisation of a case of algorithmic bias

This project is an interactive web visualization of a case of possible algorithmic bias in Germany. I did this project as the final project of the first term of my master's study program. The case is the Arabic dialect recognition algorithm for refugees, which is part of the identity management system of the BAMF and used to verify the region of origin for refugees (Biselli, 2018; Bundesamt für Migration und Flüchtlinge, 2019; Bundesamt für Migration und Flüchtlinge, 2020, Chiusi et al. 2020). The decision for the case was made in an intensive research process. I chose this particular case because it is a case of a machine learning algorithm used by the German government, which has been used nationwide in Germany for three years (Bundestag-Drucksache 19/6647, 2018). Below, you can find all the references I used for the content.

The project was exhibited at "Beyond the Frame" at the 50th International Student Film Festival Sehsüchte.


I used vue and three.js to create the website.

You can find the source code and the documentation here on my github.




You can find to total list of references on the project website. For this text, I included the following references:

Chiusi, Fabio; Sarah Fischer, Sarah; Kayser-Bril, Nicolas; Spielkamp, Matthias. (2020). Automating Society Report 2020. Algorithm Watch.

Biselli, Anna. (2018, August 20). Software des BAMF bringt Menschen in Gefahr. Vice.

Bundesamt für Migration und Flüchtlinge. (2019). Digitalisierungsagenda 2020.

Bundesamt für Migration und Flüchtlinge. (2020, October 12). Digitalisierung und Identitätsmanagement im Asylverfahren in Europa Meldung

Bundestag-Drucksache 19/6647. (2018). Antwort der Bundesregierung auf die Kleine Anfrage der Abgeordneten Ulla Jelpke, Dr. André Hahn, Gökay Akbulut, weiterer Abgeordneter und der Fraktion DIE LINKE. Einsatz von IT-Assistenzsystemen im Bundesamt für Migration und Flüchtlinge.