AI app to police online hate speech tested

An artificial intelligence (AI) application has been successfully used in Finland to identify and track hate speech online in the government-sponsored "Facts Against Hate" project, reported Xinhua, quoting a report published by the Ministry of Justice on Thursday.

One of the project's goals is to pilot a hate speech tracking AI application. Based on "teaching material" compiled by researchers, the AI application uses statistical methods to identify online hate speech and then flag it to human content moderators.

An analysis of 12 million comments posted on Finnish public social media platforms last September and October has found that under two percent of those messages qualified as hate speech. The results showed that 33 percent of all Twitter posts flagged as hate speech were actually retweets (re-postings).

A total of 298,032 hate speech messages were identified, 97 percent of them on various discussion forums.

Over half of all hate speech messages analyzed with AI were contemptuous or stigmatizing a group, and a third of them included insults or other expressions of hate.