The study's entire methodology was to look through enrollment data to find for-profit schools that had a higher proportion of black students, and then run ads for those schools.
Because a higher percentage of black people already go to the school, and are connected on Meta, and because Meta bases ad delivery on subjects relevant to people similar to / connected to you, the ads for those colleges were delivered to black people more often, because black people already are highly represented at these schools.
Ok? It's a completely generic algorithm. If the school had a uniquely high percentage of any "group" of people, whether based on race, region, hobbies, interests, literally anything, ads for that school would be more likely sent to people in those groups.
The headline is pretty editorialized and presents a pretty misleading picture.
Because a higher percentage of black people already go to the school, and are connected on Meta, and because Meta bases ad delivery on subjects relevant to people similar to / connected to you, the ads for those colleges were delivered to black people more often, because black people already are highly represented at these schools.
Ok? It's a completely generic algorithm. If the school had a uniquely high percentage of any "group" of people, whether based on race, region, hobbies, interests, literally anything, ads for that school would be more likely sent to people in those groups.
The headline is pretty editorialized and presents a pretty misleading picture.