LogFAQs > #895952546

s0nicfan posted...

Ehh... that's fair. Although I think you're under-emphasizing the importance of "sabotage" in classification, since ANY classification algorithm that uses DL or NN would be susceptible to abuse. I was speaking more from experience where its getting harder to sell customers on ML as a solution because of stories like those, and if the money dries up so does the research in that area.

I just don't think your concern about "feminism in ML" is well founded. We have laws against certain types of discrimination in this country, so it makes sense that we don't give companies carte blanche to say "I'm not being discriminatory, the blind algorithm is!" for making decisions that we wouldn't let a human make.

I don't think it really applies to the detriment of Uber here. I'm just saying, if your takeaway from this is "See, I guess there isn't structural bias in our society" but if the reason women are working less is because of societal pressures to be the person who "gives up their career" to take care of children or if they are leaving the service because of sexual harassment by customers, that doesn't erase bias, it just means it isn't at the feet of a sexist boss.