Hacker News Reader

Comment by uberman | Hacker News Reader

We are conflating problems with non representative training data and the demographics of ai practitioners.

No we're not. Quite frankly, given how many times already science and technology has failed to work for all kinds of people given the narrow view of the people that practice and create it, this kind of response is incredibly flippant.

Just to name a few:

* heart attack research failing to distinguish between different warning signs for males and females

* sensors failing to detect darker skin

* sensors at TSA being unable to scan the hair of black people because it's thicker

I agree with you but just one point - I didn't know African hair is thicker. Do you mean that African's tend to have more hair follicles on their head, or do you mean that the hairs themselves are thicker? I'm not sure if either is true. Do you have a reference.

Not sure if they mean their hair styles tend to be thicker or if their hair itself is thicker.

https://www.propublica.org/article/tsa-not-discriminating-ag...

Are you saying they are not related? When Apple release the Apple Watch, they pushed the health features. However, they didn't think to account for women's menstural cycles and how it can affect physiological indicators.

It would never occur to a guy that that type of data is missing, because it is not something he has to deal with on a monthly basis.

So, yes, the two issues are related.

I think its being conflated because the author thinks they are one in the same and is trying to make the point that they are. I happen to agree with her. However, thank you for the classic white male deflection defense.

Reply on news.ycombinator.com