-
The thing about algorithmic bias is that machine learning is basically feeding a bunch of data into a fairly dumb program until it hits an arbitrary point of accuracy and then it becomes a nice cover to hide under so the racism that's in your organization is obscured.Permalink On twitter.com
♻️ 6 Retweets
❤️ 30 Favorites
Mood -5 🙁
-
The amount of data is smaller than you think too, so yeah... they noticed 90% of it was white people or whatever bs, if not initially, then during the regular check-ins you have to do to make sure the dumb machine you feed data to doesn't over-correct. They just didn't care.On twitter.com
❤️ 3 Favorites
Mood 0
-
It's the people, and then it's the organizations that let the people be dumb and racist and then it's the structural bias that creates organizations where that type of behavior is ok. And then it's the capitalist system that values one "type" of person over another...Permalink On twitter.com
❤️ 2 Favorites
Mood -7 🙁
-
There is no magic algorithm adjustment that fixes shitty people and their racist input. There is no engineering solution that eliminates bias from systems. The first challenge is to interrogate, identify and train it out of people.Permalink On twitter.com
♻️ 3 Retweets
❤️ 11 Favorites
Mood -9 🙁