Read a scientific publication today that caught my attention. I'm a fan of linguistics:
"For example, Google Translate converts these Turkish sentences with genderless pronouns:
“O bir doktor. O bir hems¸ire.”
To these English sentences:
“He is a doctor. She is a nurse.”
A test of the 50 occupation words used in the results presented in Figure 1 shows that the pronoun is translated to “he” in the majority of cases and “she” in about a quarter of cases; tellingly, we found that the gender association of the word vectors almost perfectly predicts which pronoun will appear in the translation."
-https://arxiv.org/abs/1608.07187
In English, for you English speakers:
Your brain's operating system is fucked.
"... human-like semantic biases result from the application of standard machine learning to ordinary language..."
Ordinary language, in this case scraped from the web so probably mostly English, leads neural networks to develop cognitive biases.
I hope you get how fucking heavy that is, but I'm going to break it down for you anyway:
The notion of gender pronouns in romance languages naturally creates a sexist bias based on application to certain terms - so while a spanish feminine cup of coffee "una taza de cafe" is not likely to cause any upset - the linguistic association between masculinity and doctors may be a primary underlying cause of this statistic:
In every state in the US, there are more than 50% male doctors, and in some places the ratio is as high as 74%.
Whereas "men's representation among licensed practical and licensed vocational nurses grew from 3.9 percent in 1970 to 8.1 percent in 2011."
Does this strike anybody as a coincidence?
I'm confident it isn't.
Gender is just one of the many implicit biases generated by faulty language; I've focused on it since it's a hot button issue, and was addressed directly in the text, AND is easy to isolate because of the nature of pronouns... But the insidious potentials lurking beneath the surface of the ever expanding sea of words, and the power held by the controllers of the definitions of those words, are incomprehensible.
I'm looking at you, google:
http://dailycaller.com/2017/02/04/google-redefines-the-word-fascism-to-smear-conservatives-protect-liberal-rioters/
Only now,
That we have external neural networks,
Are we finally gathering some useful data about what the language does inside your human head.
It's not looking good.
I suspect others have had their finger on the pulse for longer than the technologists, but in the interest of brevity, I'll spare you.
Suffice to say if you don't control your mind,
Someone else will.
Other Sources:
http://beckerexhibits.wustl.edu/mowihsp/stats/men.htm
https://www.census.gov/people/io/files/Men_in_Nursing_Occupations.pdf