Facebook denies this. “At the center of these issues are false information,” Kevin McAlister said in an email. “Yes, we are a business and we make a profit, but the idea that we do this and endanger public safety or our health does not understand where our business is.”
On the other hand, the company recently criticized what the 2019 documents have disputed. “In the past, we did not address security and safety issues in the immediate future,” it said in September 2021. blog post. “Instead, we have made rapid changes in response to some of the abuses. But we have changed the way we do today. ”McAlister cited the Live Audio Rooms, which was launched this year, as an example of something made this way.
If true, that’s a good thing. The same claims made by Facebook over the years, however, will not always contradict the review. If the company is convinced of the new approach, it should start further training.
Your AI Can’t Fix Anything
On Facebook and Instagram, the importance of a post, group, or page provided to you is determined primarily by how you can view, like, comment, or share. The greater the potential, the more the platform will make the contents of your food.
But what attracts the attention of the public is this in contrast what they irritate or mislead them. This helps to explain why lowly, angry, and unassuming publishers do well on the platform. One of the internal publications, starting in September 2020, claims that “The lowest pages” find most of their followers through the News Feed concept. Another recounts the 2019 experiment in which Facebook researchers created a fake account, named Carol, and followed up with Donald Trump and a few careful publishers. Within days the tower encouraged Carol to join QAnon teams.
Facebook is aware of this dynamic. Zuckerberg himself explained 2018 the contents become more affected as they approach the violation of platform rules. But instead of rethinking the wisdom of planning to participate, Facebook’s solution has been mainly to combine a mix of censuses and machine learning to find the negatives and eliminate them. Its AI tools are widely regarded around the world; and February blog post Senior technical secretary Mike Schroepfer said, for the past three months of 2020, “97% of hate speech removed from Facebook has been detected by our anonymous machine.”
Internal notes, however, give an unpleasant picture. A conference from April 2020 stated that the removal of Facebook reduces the incidence of violence by about 19%, nudity and pornography by about 17%, and hate speech by about 1%. File from March 2021, previously reported by The Wall Street Journal, he is always skeptical. In it, industry analysts say “that we can act a little more than 3-5% hate and ~ 0.6% of [violence and incitement] on Facebook, although it is the best in the world. “