Mark Zuckerberg. Foto: Erin Scott/Reuters/Scanpix
Facebook har rekruttert kinesere fra fastlands-Kina til å delta i sensuren av hatespeech på nett. Det har en tidligere ansatt avlørt overfor New York Post. Kineserne er importert på arbeidsinnvandringsvisa. Kina driver verdens mest overvåkede samfunn. Facebook syes åpenbart de har noe å lære.
There are at least half a dozen “Chinese nationals who are working on censorship,” a former Facebook insider told me last week. “So at some point, they [Facebook bosses] thought, ‘Hey, we’re going to get them H-1B visas so they can do this work.’ ”
Kineserne arbeider fra Seattle. Varsleren hadde med seg en liste over de ansatte og håndboken i «Hatefull-tale ingeniørkunst». Teknikerne skal lære maskinene hvordan de gjenkjenner hatefull tale.
The insider shared an internal directory of the team that does much of this work. It’s called Hate-Speech Engineering (George Orwell, call your office), and most of its members are based at Facebook’s offices in Seattle. Many have Ph.D.s, and their work is extremely complex, involving machine learning — teaching “computers how to learn and act without being explicitly programmed,” as the techy Web site DeepAI.org puts it.
Algoritmene finslipes slik at konservativt innhold nedgraderes og det ønskede innhold havner på toppen. Det skjer subtilt og brukeren skal ikke merke at han blir manipulert.
It all makes for perhaps the most chillingly sophisticated censorship mechanism in human history. “What they don’t do is ban a specific pro-Trump hashtag,” says the ex-insider. Instead, “content that is a little too conservative, they will down-rank. You can’t tell it’s censored.”
Facebook er fortsatt forbudt i Kina, men det hindrer ikke Facebook å importere kinesere som kan lære bort sensur og manipulasjonsteknikker.
The Hate-Speech Engineering team’s staff includes a research scientist based at the Seattle office who earned his master’s degree in computer engineering from the Chinese Academy of Sciences in Beijing, according to his LinkedIn profile.