AI Trained to Combat Hate Speech

September 26, 2019

Professor Elizabeth Belding and Professor William Wang's research featured on two popular tech media outlets for their work on A Benchmark Dataset for Learning to Intervene in Online Hate Speech

Now, researchers from Intel AI and University of California at Santa Barbara have introduced a new generative hate speech intervention model, along with two large-scale fully-labeled hate speech datasets collected from Reddit and Gab.

Read more at Synced and TNW

 

Related Links

The Next Web , Synced Review