Facebook Explains Why Time Delay Won’t Work For Live Video And Its AI Failure

Harin


Facebook has rejected the idea that putting a time delay on live videos would stop harmful content and explained the failure of its AI technology.

In a statement responding to suggestions on adding time delay for Facebook’s live feature, Guy Rosen, VP of integrity explained why it is not going to be effective.

Time delays are usually used by broadcasters in live television to censor inappropriate language or handle unexpected situations that may happen when filming.

But Facebook said that this would not be suitable for its platform. Rosen gave several reasons for this:

- Every day, millions of live videos appear on Facebook. Even when a time delay is used, Facebook won’t be able to respond immediately. Rosen said:

- The time delay, he explained, would “further slowdown” users’ reports and content moderators’ review of inappropriate videos.

- Finally, Rosen said it would also affect first responders like law enforcement and emergency services by delaying the time they are alerted.

In the same blog post, Facebook gave further details on how its AI failed to catch the New Zealand mass shooting’s live stream.

The largest social network has been receiving criticism on its AI technology.

According to Facebook, its AI was not prepared for such kind of event, like the mosque shooting in New Zealand.

Rosen said AI has to get training on what something is and is not before the system can recognize that thing. To put it simply, if you want AI to identify terrorism, you will need to teach the system by using thousands of images of terrorist propaganda.

However, the New Zealand live stream did not trigger the system since it is a rare event.

He said:

The challenge for Facebook right now is to keep other edited versions of the shooting videos from being re-uploaded to the platform. And this is a difficult task, not just for Facebook, but for other social platforms, like Twitter, Google, and Youtube as well.

Next Story