Facebook actively fights against unacceptable content on the pages of its social network. In this struggle, the employees of the company are actively assisted by artificial intelligence. However, artificial intelligence does not perform well with all tasks, Mark Zuckerberg, CEO of Facebook, told analysts during a discussion of the quarterly performance of the company.
The involvement of artificial intelligence in moderation of content on Facebook is constantly growing. Nevertheless, you still can not completely rely on technology. Mark Zuckerberg believes that artificial intelligence is not good in everything. In particular, there are great problems with understanding human speech.
In Facebook, publications that are associated with the propagation of terrorism are very effectively deleted. According to the CEO of the social network, this is the merit of AI. In addition, artificial intelligence finds extremely effective photos of naked people. But insults and angry publications are a weak point of technology.
“It's easier to teach artificial intelligence to find nipples on images than to recognize insults,” says Mark Zuckerberg. With the help of AI, 99% of publications propagating terrorism are deleted without the participation of company employees, but it may take years to achieve effective search for insults, the founder of Facebook believes.