https://www.newscientist.com/article/2170650-anti-swearing-ai-takes-the-edge-off-abuse-on-reddit-and-twitter/

To create the system, the team first harvested just under ten million posts from Twitter and Reddit, labelled as either containing offensive language or not.

They then made an artificial intelligence algorithm made up of several parts. One part parsed a …