Facebook is using A.I. to set itself up for the next decade — and to make some cool new live video filters available to everyone using its mobile apps.
The company announced Tuesday that it has developed a deep-learning system that runs entirely on smartphones. It’s called Caffe2Go, and while it’s currently being used to make people’s videos look like they were painted by Picasso, the tech is integral to Facebook’s plans to make A.I. more accessible to developers.
“As we move forward, you can imagine how having on-device A.I. running in real time could help make the world more open and connected for people in areas like accessibility, education, or others,” Facebook researchers Yangqing Jia and Peter Vajda write in a post on Facebook’s Code blog. “The smart devices in our hands are going to continue disrupting the way we think about intelligence.”
Here’s a video uploaded by Facebook that shows it in action:
The hope is to enable more A.I. with devices that people carry with them instead of requiring massive data centers or internet connections. This could, in turn, allow developers to make smarter chatbots, release new apps on top of Facebook, and, more broadly, help the average person become more comfortable with A.I. in their everyday life. It will also help Facebook realize Mark Zuckerberg’s plans for its next decade.
Zuckerberg said during a November 2 earnings call that Facebook has three core focuses for the next 10 years: “connectivity initiatives that bring more people online, artificial intelligence, and virtual reality and augmented reality.” It just so happens that A.I. is going to prove essential to all of those projects.
Facebook chief technology officer Mike Schroepfer said Tuesday that the company is using A.I. to inform its work on connectivity, VR, and “build infrastructure that anyone at Facebook can use to build new products and services.” The future of Facebook depends on this A.I. research.
That isn’t just true for Facebook; it applies to many tech companies. Schroepfer also said that Facebook is “tackling the furthest frontiers of research, such as teaching computers to learn like humans do — by observing the world.” Researchers at Google are doing the same by helping A.I. recognize objects much like human children do.
But for now, at least, your live videos are going to get a lot more artistic.