| Aaron in for Adam today. You may have read over the past day that Facebook CEO Mark Zuckerberg apologized for the fake Nancy Pelosi video that went viral on the social network in May and maybe even that he said it was a mistake not to remove it immediately, as Google’s YouTube did. No such luck. If you watch Zuck’s lengthy answers to questions from Harvard Prof. Cass Sunstein at yesterday’s Aspen Ideas Festival event ( on the video posted here), you’ll see that’s not quite what he said. To recall, the clip relied on traditional, age-old video editing techniques, like slowing down passages, to make Pelosi appear to be slurring her words as if she was drunk. Shared on Facebook to undermine the reputation of the Speaker of the House, it was viewed over 2 million times. Pelosi blasted Facebook for not taking down the video, saying the company was “lying to the public.” At Wednesday’s event, Sunstein started off asking a basic question: “Why oughtn’t the policy be as of, say, tomorrow that if reasonable observers could not know that it’s fake, that it will be taken down and disclosure isn’t enough?” But to Zuckerberg, who has to oversee the work of some 30,000 fact and safety content reviewers, there was too much grey area in the case of the doctored Pelosi video. “We don’t think it should be against the rules to say something that happens to be false to your friends,” he began. “People get things wrong. I don’t think people would want us to be censoring that and saying that it is against the rules on this service to write something that is factually inaccurate.” Instead, Facebook’s duty is only to prevent false information from spreading quickly, he argued. If content reviewers mark something as false, “we prevent it from getting any significant amount of distribution…and we also mark it as false in the service, so anyone who sees that content sees that the content is marked as false and we show related content that is more accurate.” The mistake in the case of the Pelosi video was that Facebook’s fact checkers were too slow to tab the clip as false. “That was an execution mistake,” Zuckerberg conceded. Techie that he is, Zuckerberg had a different answer for so-called deep fakes. Those are the uncannily realistic fake videos made using artificial intelligence programs (like this one of former President Barack Obama). In those cases, maybe Facebook would take down the clips immediately, Zuckerberg said. “I definitely think there’s a good case that deep fakes are different from traditional misinformation,” he said. The company is currently reviewing its policy. The divergent answers lead to a world that’s easier for Facebook to police, but not so great for society. All I need to do is gin up my bogus video using old-fashioned techniques and it stays up, but if I use A.I. it gets taken down? I think the Russians might be able to figure that one out. There ought to be a better way, perhaps with a rapid escalation process from frontline content reviewers to a more discerning panel of experts? Earlier in the Aspen appearance, Zuckerberg called for more government regulation of the Internet, saying there were some decisions that “I don’t think people would want private companies to be making by themselves.” In the case of fake videos, that may just be the case. |
No comments:
Post a Comment