2 ms·
If you're interested in this topic, you should look at "Weapons of Math Destruction." It's a question of if the outcomes are distributed differently among race
by reureu 7y ago
If you're interested in this topic, you should look at "Weapons of Math Destruction."
It's a question of if the outcomes are distributed differently among races because the races represent truly different probabilities, or because the inputs to the model were already tainted against some race. For example, let's say you judge creditworthiness based on membership in a particular honors society, but that honors society doesn't exist at historically black colleges.
- gizmo686 7y agoIn the latter case, the solution is racism. If the non racial signals are biased against group X, then, when used in the context of those other signals, membership in group X becomes a positive signal, regardless of its independent signal. I believe there was a time when the military did precisly this for black SAT (or simmilar) score, because they found the same score tender to be smarter when it was a black person.