[link] [40 comments]
Source: [url=https://www.fark.com/comments/11845285?utm_source=feed&utm_medium=comments&utm_campaign=fark]WSJ: Facebook AI identifies and removes only 3-5% of hate speech, and 1% of all content violating its rules against violence and incitement. Facebook: Nah, we're doing great. You just have the wrong definition of *fingerquotes* 'vIoLeNt HaTe SpEeCh' [Awkward][/url]