Diverse datasets do not guarantee discrimination-free algorithms. Although a wide range of data can help improve the accuracy and fairness of an algorithm, there is still the risk of bias and discrimination. These can occur due to errors in the data, unequal distribution, or preferences in data collection. Therefore, it is essential to carefully review and monitor the datasets to ensure that they do not contain any discriminatory patterns or biases. Paola Lopez will discuss the bias in different contexts in data-based algorithmic systems and artificial intelligence. She will address various forms of bias and clarify why a conceptual differentiation is practical and structural disadvantages cannot necessarily be solved with diverse datasets and automation processes.
Paola Lopez is a trained mathematician and research associate at the Institute for Legal Philosophy at the University of Vienna. In her interdisciplinary dissertation, she investigates questions of (in)justice in the state's use of AI systems concerning individuals. In addition to her dissertation, she has developed a typology of bias forms in data-based systems, published the first analysis of the Austrian AMS algorithm and its potential for discrimination, and examined the bias discourse in the context of Twitter's image-cropping algorithm.
Moderation: Francesca Schmidt (BpB, Department of Political Education and Plural Democracy) and Ann-Kathrin Koster (Weizenbaum Institute for the Networked Society - The German Internet Institute)
The series Digitalization: Feminist & Decolonial opens up power-critical perspectives on digital transformation and transformation processes and asks, among other things, about their effects on marginalized communities.
Bundeszentrale für politische Bildung
Journalists please contact the
The Zoom access link will be sent shortly before the event on 20.10. and 24.10.
Participation in the event is only possible by visiting the provider Zoom's page. Creating an account is not necessary. The privacy notices listed there apply Externer Link: https://zoom.us/de-de/privacy.html. For more information on data protection at the bpb and your rights as a data subject, see section E. VII. Registration for digital events and online conferences of the bpb at