Misinformation made YouTube to revert back to human moderators

Google’s YouTube has reverted to utilizing extra human moderators to vet dangerous content material after the machines it relied on throughout lockdown proved to be overzealous censors of its video platform.

In March, YouTube said it would rely more on machine learning systems to flag and remove content that violated its policies on things like hate speech and misinformation. But YouTube told the Financial Times this week that the greater use of AI moderation had led to a significant increase in video removals and incorrect takedowns.

Around 11 million videos were removed from YouTube between April and June, says the FT, or about double the usual rate. Around 320,000 of these takedowns were appealed, and half of the appealed videos were reinstated. Again, the FT says that’s roughly double the usual figure: a sign that the AI systems were over-zealous in their attempts to spot harmful content.

“WE WERE GOING TO ERR ON THE SIDE OF MAKING SURE THAT OUR USERS WERE PROTECTED”

YouTube’s chief product officer, Neal Mohan, told the FT: “One of the decisions we made when it came to machines who couldn’t be as precise as humans, we were going to err on the side of making sure that our users were protected, even though that might have resulted in slightly higher number of videos coming down.”

Even with more straightforward decisions, machines can still mess up. Back in May, for example, YouTube admitted that it was automatically deleting comments containing certain phrases critical of the Chinese Communist Party (CCP). The company later blamed an “error with our enforcement systems” for the mistakes.

Amid widespread anti-racism protests and a polarising US election marketing campaign, social media teams have come beneath rising strain to raised police their platforms for poisonous content material. Specifically, YouTube, Fb and Twitter have been updating their insurance policies and know-how to stem the rising tide of election-related misinformation, and to forestall hate teams from stoking racial tensions and inciting violence.

The pace at which machines can act in addressing dangerous content material is invaluable, stated Mr Mohan. “Over 50 per cent of these 11m movies had been eliminated with out a single view by an precise YouTube consumer and over 80 per cent had been eliminated with lower than 10 views. And in order that’s the ability of machines,” he stated.