4 ms·

> The presumption is that this is racism, but is there any support that it’s not caused by physiological differences? The conclusion is that the system is bias

> The presumption is that this is racism, but is there any support that it’s not caused by physiological differences? The conclusion is that the system is biased towards a particular race. The reasons WHY it was biased makes no difference to the fact that the system IS biased.

FakeComments 8y ago
The NBA and NFL are biased against white people, by that standard. Why is everyone up in arms about a small deployment of a new technology instead of national networks with hundreds of millions of fans being biased? Because this is race baiting meant to portray response to genuine physiological differences as racism — just like my usage of the NFL and NBA above was.
- sctb 8y ago
  Please stop. > Please don't use Hacker News primarily for political or ideological battle. This destroys intellectual curiosity, and we ban accounts that do it. > Eschew flamebait. Don't introduce flamewar topics unless you have something genuinely new to say. Avoid unrelated controversies and generic tangents. https://news.ycombinator.com/newsguidelines.html https://news.ycombinator.com/newsguidelines.html
SlowRobotAhead 8y ago
>The reasons WHY it was biased makes no difference to the fact that the system IS biased. Disagree. If some process or algorithm gives biased results, there is a BIG difference if it’s because of a technical limitation like “darker people are harder to see” or incompetence like “we didn’t test enough dark people” or maliciousness as “we just don’t like dark people”. So it does make a difference, esp when considering if it’s worth firing the the ol’ faux outrage machines.
james_s_tayler 8y ago
It might do? There are 5x more white people than black people in the US (roughly). So... it's kind of understandable that any training data set would contain more white people than black people if you didn't filter it beforehand? I guess theoretically you would have to limit the size of the training data for all groups to the size of the set of training data you have on the smallest group, so that all groups could be represented equally in the training data, but then perhaps you get a far less accurate model overall and it results in greater negative consequences? Just spit balling here. But it's a complex problem. The whole 'racist AI' thing screams Conway's Law to me.
- thereisnospork 8y ago
  If anything, shouldn't groups be represented proportionetly? Else the lack of viable inuit training photos is going to put a real cramp on everything.

FakeComments 8y ago

The NBA and NFL are biased against white people, by that standard. Why is everyone up in arms about a small deployment of a new technology instead of national networks with hundreds of millions of fans being biased? Because this is race baiting meant to portray response to genuine physiological differences as racism — just like my usage of the NFL and NBA above was.

sctb 8y ago
Please stop. > Please don't use Hacker News primarily for political or ideological battle. This destroys intellectual curiosity, and we ban accounts that do it. > Eschew flamebait. Don't introduce flamewar topics unless you have something genuinely new to say. Avoid unrelated controversies and generic tangents. https://news.ycombinator.com/newsguidelines.html https://news.ycombinator.com/newsguidelines.html

SlowRobotAhead 8y ago

>The reasons WHY it was biased makes no difference to the fact that the system IS biased. Disagree. If some process or algorithm gives biased results, there is a BIG difference if it’s because of a technical limitation like “darker people are harder to see” or incompetence like “we didn’t test enough dark people” or maliciousness as “we just don’t like dark people”. So it does make a difference, esp when considering if it’s worth firing the the ol’ faux outrage machines.

james_s_tayler 8y ago

It might do? There are 5x more white people than black people in the US (roughly). So... it's kind of understandable that any training data set would contain more white people than black people if you didn't filter it beforehand? I guess theoretically you would have to limit the size of the training data for all groups to the size of the set of training data you have on the smallest group, so that all groups could be represented equally in the training data, but then perhaps you get a far less accurate model overall and it results in greater negative consequences? Just spit balling here. But it's a complex problem. The whole 'racist AI' thing screams Conway's Law to me.

thereisnospork 8y ago
If anything, shouldn't groups be represented proportionetly? Else the lack of viable inuit training photos is going to put a real cramp on everything.