AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
How to Use Toxigen to Fight Toxic Language
The focus is on creating a dataset of those kinds of harmful things so that it can be leveraged as research. And they found based on human evaluators that these are pretty realistic. It's hard to differentiate from the sorts of people. Little nasty people on the internet could say. So yeah, very cool work and interesting approach here.