قالب وردپرس درنا توس
Home / Technology / Facebook builds AI watchdog chips to filter live video

Facebook builds AI watchdog chips to filter live video



Facebook develops its own chips, with which video content can be analyzed and filtered in real time. This clears the way to replace human moderators as streaming in the social network grows exponentially. The site has been criticized in recent years for allowing users to share controversial or offensive content, including shots of people who have been attacked or committed suicide.

Facebook currently relies on a combination of automated rating and filtering and a human team of moderators. The latter work around the clock, responding to content tagged by Facebook users, as well as what has surfaced through the automated systems. However, it is unclear how scalable a solution is.

As a result Facebook began to look for alternatives. This includes working on a self-developed chip design that is more effective and energy efficient in analyzing and filtering live video. The project was announced by Yann LeCun, Chief Artificial Intelligence Scientist at Facebook, Bloomberg, at the Viva Technology Conference in Paris.

The potential for Facebook is clear. "Imagine, someone uses Facebook Live to film his own suicide or murder," LeCun theorized. "They want to be able to put off this kind of content just as it is happening."

The idea is to use a dedicated AI chip that would focus on searching videos and real-time evaluating what the content is showing. While this kind of analysis is not impossible with today's hardware, it typically requires significant amounts of processing power. This has a serious impact on energy consumption and costs.

"There is a great deal of effort being put into developing chips that are more energy efficient," LeCun confirmed. "Many companies are working on this, including Facebook." The company already designs its own data center hardware, emphasized LeCun.

It's not the first time we've heard the babble of Facebook internally looking for customized silicon. Already in April, reports had shown that the social network followed in the footsteps of Apple, Google and others and developed its own chipsets. Several job ads for new roles focused on processors were discovered, covering a variety of topics.

At F8 201

8, the company's annual developer event, AI, took center stage. This included a new set of video analysis tools, in addition to AI specialized in translation.


Source link