【谷歌多手段抑制极端主义内容传播】

【谷歌多手段抑制极端主义内容传播】谷歌最近一直面临着打压极端主义内容的压力。起初,他们利用机器学习制造自动过滤器,并在疑似恐怖主义视频中投放反恐广告,而后又雇佣了大量员工审查YouTube视频,采用给可疑视频添加警示标语等方式平衡言论自由与公众安全,并去除了恐怖主义视频中的商业广告。

Google bets AI and human oversight will curb online extremism

0930

Google is under a lot of pressure to stamp out extremists' online presences, and it's responding to that heat today. The internet giant hasoutlined four steps it's taking to flag and remove pro-terrorism content on its pages, particularly on YouTube. Technological improvements play a role, of course, but the company is also counting on a human element that will catch what its automated filters can't.

To start, it's pouring more energy into machine learning research that could improve its ability to automatically flag and remove terrorist videos while keeping innocently-posted clips (say, news reports) online. It's also expanding its counter-radicalization system, which shows anti-extremist ads to would-be terrorist recruits.

It's the stronger reliance on people that may matter the most, however. Google plans to "greatly increase" the number of humans in its YouTube Trusted Flagger program, improving the chances that it'll catch terrorist material. It's likewise working with anti-extremism groups to pinpoint recruiting-oriented content. Google wants to tackle those YouTube videos that are borderline, too -- if it spots videos with "inflammatory" religious or supremacist material, it'll put those clips behind a warning and prevent them from getting ad revenue, comments or viewing recommendations. In theory, this strikes a balance between free speech and public safety.

To some extent, the plans are an extension of Google's ongoing efforts, such as its plan to pull ads from extremist videos. Still, they might just assuage politicians who have threatened to institute legal mandates for anti-extremist takedowns. Google, Facebook, Twitter and others have already stepped up their collective fight against terrorism, but this is a relatively concrete roadmap. The big question is whether or not all these initiatives will be enough. AI-powered flagging and greater oversight could help, but the sheer volume of videos on YouTube makes it entirely possible that some footage will slip through the cracks.

https://www.engadget.com/2017/06/18/google-bets-on-ai-and-human-oversight-to-curb-online-extremism/


Comments are closed.



无觅相关文章插件