AI Ethics: Fixing it?
With the great current breakthroughs in AI and its deployment come great concerns about AI’s impact on society and the planet. These concerns imply many responsibilities, opportunities for action, and frustrations for different stakeholders. Computer scientists and other academically-educated professionals (i.e. 'us' as the presenters and audience of this talk) form one, or even several, of these stakeholder groups. How can we have a principled discussion about our responsibilities?
Praktische info:
georganiseerd door:
After an introductory example, in this talk, we will look into professional ethics codes for guidance, and ask how to make principled choices regarding the values that are often alluded to only loosely in these codes. We then focus on the value of reducing inequalities and avoiding discrimination. We take a critical look at 'de-biasing', a term for an evolving set of computational methods that have received a lot of attention as potential answers to AI challenges from within the AI community. We open the discussion to explore how researchers’ action spaces can be enriched through interdisciplinary collaboration. We also ask how this focus on individual responsibilities can and must be complemented by structural adjustments. Last but not least, we aim at furthering participation opportunities through technology by organising this talk as a hybrid live event.
Lesgevers / sprekers
Bettina Berendt holds the Chair for Internet and Society at Technical University of Berlin. She is also a director of the Weizenbaum Institute and a visiting professor at KU Leuven, Belgium. Her research includes Data Science and Critical Data Science, especially with respect to Privacy/Data Protection, discrimination, and fairness, as well as AI and ethics, with a focus on textual and web-related data. Further details can be found on her page.
Kristen M. Scott is a PhD candidate in the department of Computer Science at KU Leuven University in Leuven, Belgium, working under the supervision of Prof. Bettina Berendt. She is an Early Stage Researcher in the NoBias project, a Marie Curie ITN. Her work focuses on the topic of bias and fairness in algorithmic decision-making, particularly on biased representations of people in data. Kristen’s specific interest is in the question of how to effectively involve all stakeholders and impacted persons in the design of algorithmic systems in the interest of addressing algorithmic harms.
