YouTube Bots Can Now Flag Objectionable Content Before Being Viewed

According to Google’s transparency report, over the last quarter, human moderators and machine learning algorithms took down 8,284,039 YouTube videos that violated the company’s terms of use and policies. The videos violated the website policies about adult content, harassment, violence, and hate speech.

Of the videos taken down, 6.7 million were flagged by computer algorithms before somebody was even able to watch them. The report also shows that 1.5 million controversial videos were posted on the site but were quickly removed.

The Google-owned YouTube boasts that its machine learning algorithms are getting smarter and could soon flag banable videos without any human input. The company now heavily relies on bots to weed out objectionable content because there is no way human moderators can inspect millions of clips in such a short time.

Over last year’s Q4, controversial videos were being removed at a rate of one video per second.

YouTube Bots Are Getting Better at Their Job

The report was made public during an earnings call on Monday. Google CEO Sundar Pichai noted that 6 million videos were scrapped from the video streaming site in the fourth quarter alone. Of those, 75% of the videos disappeared before anyone had a chance of watching them.

Pichai praised the site’s machine learning systems for the feat. YouTube confirmed in a recent blog post that more than 8 million videos were removed from the platform in Q4. Most of the videos were spam or adult content.

YouTube bots are getting better at spotting clips that promote violent extremism, the site announced. Last year, eight percent of those videos were removed before they could be seen ten time. This year, more than half of these videos are weeded out before they get 10 views, according to the company.
Image Source: Maxpixel