Racism – A statistical comment

Big data and artificial intelligence may be useful in fighting racism, if we replace race as a (weak) predictor of future behaviour (eg past of future crime) [Reference missing] with other (sufficiently) informative predictors (eg social score?). However, this would require citizens to share personal information with public authorities, which creates new risks. However, by sharing personal information citizens could provide protection for other citizens which are discriminated against to render race uninformative.

A similar approach could be applied to other sensitive predictors. For example, a gender-based evaluation of a job application is probably only predictive of future job performance (depending on the domain) if no other information is available. This means, gender may be a relevant predictor before reading an applicant’s CV but irrelevant after reading the CV. Discrimination, prejudice or heuristic-based assessments [1] are usually applied if little information is available, so naturally providing information will help fight discrimination (in theory).

[1] https://en.wikipedia.org/wiki/Heuristic

http://wilmarigl.de

de_DE_formalGerman