The seminar will feature six lectures. We have decided to also post the slides of the presentations online and also a very small excpert from the discussions. There will also be a final submission by every participant, which will be posted here. This lecture has dealt mainly with two topics:
- What is an Algorithm?
First we started with a short presentation: Critical Algorithm Studies – Introduction
Then we discussed and reflected upon different definitions of Algorithm. Technological definitions focus on tool-perspective and socio-technical definitions acknowledge people using, developing, thinking about, being influenced by Algorithms. Usually Algorithms are obfuscated and almost mystical, which is maybe not clear to many computer scientists. We came to the conclusion that we do not really need an all encompassing definition of Algorithm.
- Embedded Values and Biases of Algorithms
Again a short presentation was given: Embedded Values and Biases
The first part of the discussion was focused on responsibility. Then we talked about good practises and how to reduce “bias”. We came to the conclusion that models inherently have bias, since they can not completly represent reality and are done by people with a goal in mind. The goal should be to be aware of biases and then to ethically “balance” them out. The paper on facial recognition showed that the researchers were not aware of the bias in data and only indipendent tests outside the lab showed them the problematic results, which in turn means participatory approaches need to be considered in algorithm design.