On a related note, I find it rather awkward that software, and online services are allowed to discriminate by applying automated classifiers to datasets.
For instance, an insurance company could deduce from the fact that a person likes black music, that he is 2x more likely to be involved in an incident of some kind.
To me, it seems that this is a form of prejudice and discrimination, that, in real life, would not be accepted. So why do we accept it from these big companies?
And here comes the main point: this should apply even if "black music" is replaced by some trait that doesn't touch upon racism, and the company is not an insurance company but, for instance, an advertisement agency.
Comments
On a related note, I find it rather awkward that software, and online services are allowed to discriminate by applying automated classifiers to datasets.
For instance, an insurance company could deduce from the fact that a person likes black music, that he is 2x more likely to be involved in an incident of some kind.
To me, it seems that this is a form of prejudice and discrimination, that, in real life, would not be accepted. So why do we accept it from these big companies?
And here comes the main point: this should apply even if "black music" is replaced by some trait that doesn't touch upon racism, and the company is not an insurance company but, for instance, an advertisement agency.