[peoplewindow]

Well, according to the notes there was an entire talk about bias where the example of a Turkish sentence that didn't contain gendered pronouns "X is a nurse, X is a doctor" gets translated to "She is a nurse, he is a doctor" by Google Translate. And apparently that's a problem that needs fixing rather than a reasonable translation, although we all know that Google Translate is trained on the work of human translators, where it presumably learned this (demographically correct) bias from. So I guess they need fixing too.

Seems like NIPS might have a systematic issue with hyper-sensitivity to perceived slights. But it's the intersection of academia and SV companies so is it really a surprise?