These scientists are so precious: <\/p>\n\n\n\n
<\/p>\n\n\n\n
“One might have hoped that the Google News embedding would exhibit little gender bias because many of its authors are professional journalists,\u201d they say.<\/p>\n\n\n\n
<\/p>\n\n\n\n
Aha. Ahahaha<\/em>. <\/p>\n\n\n\n <\/p>\n\n\n\n Ahem. <\/p>\n\n\n\n <\/p>\n\n\n\n This post could also have gone in my SFF Thought Starters collection. There’s a great story in the Futuristica<\/em> anthology about an AI cop that’s shooting young black men because it’s been trained to assess threats based on a corpus of previous police interactions. <\/p>\n\n\n\n <\/p>\n\n\n\n Originally shared by Winchell Chung<\/strong><\/p>\n\n\n\n <\/p>\n\n\n\n And how the MIT researchers used a mathematical transform to remove the odious gender bias from the dataset.<\/p>\n\n\n\n Bias example: If you query the vector space embedding asking Man is to Programmer the way Woman is to X, the dataset will respond “Homemaker”.<\/p>\n\n\n\n