Flagging offensive content can be pretty hard. And when you are attempting to do it on a platform of Facebook’s magnitude — the task becomes herculean. And when you need to supervise live video streams as well — Well, its a superhuman task. Which is why Facebook is now turning to AI to help it monitor offensive material in live video streams.
Up until now, the company has relied mostly upon user reports to flag and remove offensive content. However, that is set to change as the company slowly increases it reliance on AI to help it combat such material. The company has apparently developed and is using an an algorithm that is able to detect nudity, violence, or any of the things that are not according to the company’s policies.
The system had been activated to flag all those extremist videos that briefly occupied a lot of space on the social networking platform earlier this year. However, the company is now deploying it to monitor live-streams as well.
While speaking with Reuters, Joaquin Candela, Facebook’s director of applied machine learning said,
One, your computer vision algorithm has to be fast, and I think we can push there, and the other one is you need to prioritize things in the right way so that a human looks at it, an expert who understands our policies, and takes it down.
And that is not all. The company is bringing AI almost everywhere — for instance, using it to automate the process of sifting through all the tens of millions of reports the social media platform receives and sorting duplicate ones in a separate pile.
Meanwhile, there are still some major roadblocks that will need to be crossed before an AI — or a human for that matter — can began to sift through all the news that comes to Facebook. The content populates the social networking website has come to the fore recently following an outcry after reports of fake new that may have influenced US elections and after an iconic Vietnam War photo was removed — and later restored — due to nudity.
The company is working hard to sort through these issues and considering the sheer magnitude of users on the platform — AI just might be its best bet.