Facebook is building chips to analyze and filter live video

Facebook is in the process of making its own chips for filtering video content, chief AI scientist Yann LeCun revealed today in a talk at the Viva Technology industry conference in Paris.

Conventional methods are no longer sufficient as they require too much energy and compute power, LeCun said, according to Bloomberg. Last month, Bloomberg reported that Facebook is building its own semiprocessor.

Facebook currently uses Intel CPUs for many of its AI services, engineering manager Kim Hazelwood recently told attendees at Intel AI DevCon this week. A transition to a specialized chip could help the company more quickly filter video content for violations of its terms of service, like if a person commits suicide in a livestream or performs another act of violence. Facebook currently uses various forms of computer vision to do things like remove content by terrorist organizations like ISIS.

Graphics processing units (GPU) from companies like AMD and Nvidia assist with the rapidly increasing amount of unlabeled data like images and videos in the world. Google continues to accelerate AI training and deployment with its tensor processing unit (TPU), and Intel’s Stratix 10, a field programmable gate array (FPGA) chip, is being used for Microsoft’s Project Brainwave.

Apple, Google, and Amazon are also reportedly planning to make specialized chips to speed up AI services for their hardware products.

:story: VentureBeat