Google has developed an algorithm that flags 'toxic" messages intended to drive others away from a conversation. The algorithm can be very useful for comment sections on news sites, where smart comments are usually interrupted with obscenities and other verbal attacks, usually off-topic. 

To avoid 'toxic' comments, some media outlets like The New York Times count on a team of human moderators that checks nearly every single comment before it gets published. But this human resource is too expensive to afford for the majority of the news outlets, reports The Atlantic. Thus, the algorithm developed by Google can be very useful, and cheaper too. 

The tool, called Perspective, uses machine-learning algorithms to separate out the worst comments that people leave online. 

Read the full story in The Atlantic.