Gender Discrimination on steroids via AI

The bachelor thesis deals with the discrimination potential of Machine Learning Algorithms withinDecision – Support – Systems (DSS), which are increasingly used in German companies. The specialization is in the areas of recruiting and marketing. The concrete goal of the thesis was to find out whether AI – based systems reinforce gender discrimination or not.

The elaboration shows that AI – based systems in the form of DSS can actually provide a breeding ground for gender discrimination. This does not result from the technology itself, but from a multitude of factors, such as distorted training data, the algorithms created and the involvement of a multitude of people with different competences and characteristics in the development process of such systems. Despite the still low use of AI in Germany, there is a risk of emancipatory regression in the future, as an enormous increase in use is planned. So far, no uniform framework conditions for data acquisition, transmission and processing have been established, user companies themselves have little know-how in the field of AI, there is a great lack of data specialists and action against digital gender discrimination is generally not well received.

This results in different dangers for the areas considered, which can affect the individual’s freedom of development, the right to comprehensive information and the personal freedom of decision of the individual. A holistic rethinking must therefore take place and a catalogue of criteria must be drawn up for the creation, use and long-term handling of AI – based software. A first approach is provided by a package insert for algorithmic systems as well as a code of ethics and recommendations for action for the various stakeholders developed within the bachelor thesis.