AI that knows hate: Facebook removes offensive posts before users report

The Reinforced Integrity Optimizer (RIO), unveiled on Thursday, is meant to amplify the decision-making powers of human reviewers of hate-speech posts to make the process faster.