display
Menlo Park (dpa) - Facebook is making further progress in filtering out prohibited content on the platform with the help of artificial intelligence.
In the past quarter, 97 percent of remote hate speech from software was discovered before people saw it, Facebook announced on Thursday.
In the quarter before that, the proportion was 94 percent and a year earlier only a good 80 percent.
The online network is also confident that it cannot only proceed in this way with text contributions.
"We understand audio and video, we understand the content," said technical director Mike Schroepfer.
Among other things, by who shares what, the software can also classify statements in a broader context.
display
During the corona pandemic, Facebook sometimes had problems getting the teams that check content to work just as efficiently from home as from the office.
© dpa-infocom, dpa: 210211-99-405226 / 2
Facebook on the role of artificial intelligence