New Delhi: Unraveling media reports on their inability to deal with hatred speech, Facebook now claims that the prevalence of hatred speech on its platform has dropped by almost 50 percent in the last three quarters.
The claim came in response to a report on the Wall Street Journal (WSJ) on Sunday, which said that Facebook content moderators did not succeed in eliminating content that did not deserve to use artificial intelligence (AI).
A reply, Guy Rosen, Vice President of Integrity on Facebook, said that their technology had a big impact to reduce how much people talked on Facebook.
“According to the latest community enforcement report, the prevalence is around 0.05 percent of the content seen or around five views per 10,000, down almost 50 percent in the last three quarters,” he added.
“Data drawn from the leak document is used to create a narrative that the technology we use to fight inadequate hatred speeches and that we deliberately misinterpret our progress.
This is not true,” Rosen said.
The WSJ report claimed that internal documents showed that two years ago, Facebook reduced the time that human reviewers focused on complaints of hatred speech, and making other adjustments that reduce the number of complaints.
“That in turn helped create the appearance that AI Facebook was more successful in enforcing the company’s rules than actually,” said the report.
Rosen said in postpost that focuses only on content transfer is the wrong way to see how we fight speech hatred.
“We need to be sure that something is hating speech before we delete it.
If something can be hated, but we are not confident enough so that it fulfills the blade to be removed, our technology can reduce the distribution of content or will not recommend groups, pages or people who in Regularly post content that tends to violate our policy, “he said.
Facebook said when starting to report metrics about hatred speech, only 23.6 percent of the content that was removed was detected proactively by the system; The majority of what he removed was found by people.
“Now the number is more than 97 percent.
But our proactive level does not tell us what we miss and do not take into account the number of our efforts, including what we did to reduce the distribution of troubled content,” The word executive Facebook.