top of page
Battling Discrimination in Artificial Intelligence
Prof. Tamar Kricheli-Katz & Prof. Ronen Avraham
The use of artificial intelligence in outcome prediction algorithms has been growing more prevalent over the past few decades. However, the more popular the technology becomes the more adamantly the argument is raised that, due to several different reasons, these algorithms are susceptible to bias and therefore their outputs may be discriminatory. The possibility that such algorithms can be biased has been confirmed by numerous empiric studies which span disciplines and models. In this report we present the most prominent and cutting-edge solutions for AI-based discrimination that exist in academic literature. These methods span from wholly legal solutions, through multidisciplinary ones, to social and economic approaches and finally to technical and statistical options. The different solutions are grouped into categories based on the step in the AI model process they intervene or apply to.
bottom of page