There are many examples of artificial intelligence (AI) discrimination: automatic facial recognition, for instance, responds best to faces of men with light skin. Other examples range from Amazon’s development of an algorithm to evaluate applications to job ads, which was found to discriminate against women, and an algorithm designed to calculate offenders' likelihood of recidivism, which was discovered to rate white people more positively and people of color more negatively.
Discrimination is one of the most pressing social consequences of the use of AI. Professor Dr. Tobias Matzner's talk shows that preventing discrimination through AI first requires proper understanding, as discrimination is more than simple bias in the data. The talk then delves into how the very explanations of how AI works can help prevent discrimination on the part of AI. Consistent with the definition of discrimination developed in the first part of the talk, explanations need to take into account different social contexts.
Event details
Date: Wednesday, March 30, 2022, 5pm-6 pm.
Online lecture via zoom: meeting ID: 987 0689 1523, password: 194439
Quicklink: https://uni-bielefeld.zoom.us/j/98706891523?pwd=OU1CZGNUTUVHRWtSa0luV29KTXRjZz09
Program
17:00 Welcome and introduction to the research area
Prof. Dr. Katharina Rohlfing
Speaker TRR 318
17:10 Presentation "Discrimination by AI – and how explanations can help".
Prof. Dr. Tobias Matzner
Subproject leader TRR 318
17:40 Presentation of AI workshops for secondary students, medical students and decision-makers
17:45 Discussion
The talk will be given in German.
Registration is not required. We look forward to your participation.
Contact:
Linda Thomßen
communication(at)trr318.uni-paderborn(dot)de
Phone: 0521 106-86695
