Stereotypes in Commonsense Reasoning Systems[go to master theses]
Open Master Thesis - Contact a supervisor for more details!
Commonsense Reasoning is a branch of artificial intelligence concerned with modeling the human ability to draw conclusions in everyday situations. There are a variety of benchmarks for this area that test different aspects of human reasoning. Various systems have been developed for these benchmarks, some of which achieve quite good results.
The aim of this master's thesis is to investigate the extent to which state-of-the-art systems for commonsense reasoning display human prejudices.
The master thesis includes the design of methods to measure prejudices present in commonsense reasoning systems. These methods will then be used to determine the degree of existing prejudice in state-of-the-art systems.
In addition, the reasons for stereotypes in these systems will be analyzed and initial ideas for approaches that avoid such stereotypes will be developed.