Discrimination seems to always have a negative tone nowadays, but in some contexts it is a virtue. E.g. "he has discriminating tastes". I hope any AI I make is discriminating.

For better or worse, what we accept depends on the situation. For example, health insurance discriminates on gender, race, and age, all of which are illegal to discriminate on in employment matters.

Personally, I find advertising to be in the "okay" category for racial discrimination. Advertisers essentially pay for Facebook users' data (which is why Facebook is free as in beer), and use that data to target ads. For example, some cosmetic lines cater to dark skin, and racial discrimination keeps their ads relevant.

You choose your profession, but race largely corresponds with personal qualities not of your choosing. I'd imagine this is particularly true when it comes to Facebook's inference of a person's race, since I'm pretty sure it's not user-supplied.

> Satterfield added that the “Ethnic Affinity” is not the same as race — which Facebook does not ask its members about. Facebook assigns members an “Ethnic Affinity” based on pages and posts they have liked or engaged with on Facebook.

You're right that discrimination, taken broadly, is a morally neutral concept. But the law establishes protected classes to attempt to prevent people from be disadvantaged or disenfranchised due to qualities society feels are unacceptable to discriminate by.

And there's something especially galling about the fact that this functionality excludes people of ethnic categories for targeting, rather than targeting products of particular interest. Both are pretty bad though.