The latest paper shows that some more troubling implicit biases seen in human psychology experiments are also readily acquired by algorithms. […] The AI system was more likely to associate European American names with pleasant words such as “gift” or “happy”, while African American names were more commonly associated with unpleasant words.
These biases can have a profound impact on human behaviour. One previous study showed that an identical CV is 50% more likely to result in an interview invitation if the candidate’s name is European American than if it is African American. The latest results suggest that algorithms, unless explicitly programmed to address this, will be riddled with the same social prejudices.
Nothing, absolutely nothing, has to do with race nowadays, in the eyes of white America writ large. But the obvious question is this: if we have never seen racism as a real problem, contemporary to the time in which the charges are being made, and if in all generations past we were obviously wrong to the point of mass delusion in thinking this way, what should lead us to conclude that now, at long last, we’ve become any more astute at discerning social reality than we were before? Why should we trust our own perceptions or instincts on the matter, when we have run up such an amazingly bad track record as observers of the world in which we live? In every era, black folks said they were the victims of racism and they were right. In every era, whites have said the problem was exaggerated, and we have been wrong.