MOSCOW, October 25. /TASS/. The inefficient system of moderation of political content in Middle Eastern and Asian states indicates Facebook’s technological problems and the imperfection of its artificial intelligence, says Russian Federation Council Commission on Informational Policy and Media Relations head Alexey Pushkov.
Earlier on Monday, AP reported citing the social platform’s internal documents that Facebook was unable to build an effective system of moderation of political content in Middle Eastern and Asian states due to the lack of knowledge of local languages and culture. According to the report, Facebook employees and algorithms subjected statements in Arabic to excessively harsh moderation, erroneously taking them for political calls. Thus, a large number of neutral Arabic words that could have potentially been used by various radical groups in their names or calls were blacklisted.
"The platform’s inability to separate the wheat from the chaff and piling together well-intended messages and messages that in one way or another reflect extremist views is connected, in my opinion, with Facebook’s limited technical capabilities. It is no secret now that the artificial intelligence implemented on the platform is unable to pinpoint illegal content precisely. This intelligence partially passes the illegal content and partially blocks things that cannot be illegal by definition," Pushkov said.
He emphasized that he is talking only about technological problems of recognition of illegal content. According to the senator, this is not an intentional discrimination of the Arabic world by the social platform, because similar cases have been registered "both on Russian-speaking and English-speaking platforms of this social network."
"It has become known recently that Facebook platforms that implement artificial intelligence are unable to efficiently identify and block hate speech, as well as photo and video footage that contain excessive violence, not only in Arabic countries, but all around the world," Pushkov underscored.
"Facebook will, apparently, work to improve the AI recognition capabilities. I believe they will do that, because otherwise it would weaken their image and create an impression that the company is simply not effective enough and adequate for the tasks it sets up before itself. Facebook is interested in income, in subscriber growth. I believe that they will look for some technological response to this by working on the artificial intelligence," the senator explained.