If there are two things Facebook is notoriously bad at right now, it’s maintaining the privacy of its users and developing cool Facebook-related hardware. Yet, according to a new job posting first noticed by Bloomberg, Facebook may start taking hardware a lot more seriously.
The primary job listing to note is for a manager in charge of ASIC Development. ASIC stands for Application Specific Integrated Circuit, which means exactly what it says: a processing chip intended for a specific application. Bloomberg notes this could be anything from a chip intended for future Oculus headsets to a CPU necessary for future Facebook servers, which is definitely true—but the really important thing to note is, this wasn’t the only listing for a new chip design program.
Crucially Bloomberg noticed the posting for the manager because a related job posting was first tweeted by Facebook’s head of artificial intelligence.
This suggests Facebook is about to get very seriously about chips that better handle AI. Those kind of chips could be found in something like a smart speaker, which needs an AI to understand your words, or a server, which needs an AI to process images of your face from a 10-year-old photo your brother just posted on the Facebook website.
Apple, Qualcomm, and Huawei all produce CPUs for phones that have a focus on improving AI processing. Google and Intel are also exploring it, and Nvidia is making a lot noise in the AI and machine-learning chip space and has invested billions in attempt to be the leader in this new avenue of chip design. So Facebook, coming in with little more than some job listings on its career site, appears to be late to the party.
Yet Facebook has recently touted AI as one potential tool to employ to better combat hate speech or to understand your future behavior so advertises can alter it. The company has embraced AI for some time now, and currently uses it not just to identify you and your friends in images, or create large and detailed profiles on you, but also to populate your newsfeed with news and posts it thinks you want to see. Designing its own AI chip is a smart way to keep things in house and cut down on costs, but it’s also a little terrifying that so frustratingly opaque a company could be working on the brain for Skynet right now.