Facebook is building new AI tools to help detect hate speech on the platforms. AI now proactively detects 94.7 percent of hate speech removed from Facebook, up from 80.5 percent a year ago and up from just 24 percent in 2017.
This has been driven by advances in our automated detection tools, such as deploying XLM, Facebook AI’s method of training language systems across multiple languages without relying on hand-labeled data sets. We’ve recently deployed new systems that have further improved our ability to detect hate speech. These include Reinforced Integrity Optimizer, a system that learns from real examples and metrics, and using our Linformer AI architecture, which enables us to use cutting-edge language-understanding models that were previously too big and unwieldy to work at scale.