As Facebook Inc. has expanded its scope far beyond the realm of social media in the last few years, no area is arguably more important than its recent push into artificial intelligence research. This week, Facebook Chief Technology Officer Mike Schroepfer outlined some of the company’s current research projects and the role AI plays in Facebook’s plans.

Facebook is currently focused on three main areas of research – connectivity, virtual reality and artificial intelligence – and according to Schroepfer, AI plays a role in everything Facebook is doing in each of these fields.

“Our work in AI is helping us move all these projects forward,” Schroepfer said. “We’re conducting industry-leading research to help drive advancements in AI disciplines like computer vision, language understanding and machine learning. We then use this research to build infrastructure that anyone at Facebook can use to build new products and services.”

“We’re also applying AI to help solve longer-term challenges as we push forward in the fields of connectivity and VR. And to accelerate the impact of AI, we’re tackling the furthest frontiers of research, such as teaching computers to learn like humans do — by observing the world.”

Connectivity

Facebook has quite a few connectivity projects in the works, including high-powered lasers that can transmit data across long distances with no cables required. These lasers, along with other connectivity technologies, will eventually make their way out into the world on Facebook’s massive Aquila drone (above), a solar-powered autonomous aircraft that is designed to fly continuously for months at a time without ever landing.

Aside from the the systems that keep Aquila up in the air, there are other connectivity projects that take advantage of AI, including Terragraph, a computer vision program that creates 3D city maps that can be used to better plan new infrastructure.

“As wireless networks become denser with increasing bandwidth demand, this automated solution lets us process more radio installation sites at a finer granularity,” Schroepfer explained.

Virtual Reality

Facebook is currently one of the biggest names in the latest virtual reality boom thanks to its $2 billion purchase of Oculus VR LLC. in 2014. Oculus is the creator of both the Oculus Rift and Samsung Gear VR headsets, and Facebook has been working not only to improve the hardware, but also to create new software to power all sorts of VR experiences, including 360-degree video.

In August, for example, Facebook announced a new stabilization technology that uses computer vision to smooth out shaky 360 video. This is important because while regular shaky video could make viewers feel a little motion sick, shaky 360 video could make viewers “violently ill,” to borrow a phrase from Oculus CTO John Carmack.

Aside from computer vision, Facebook has also been working to incorporate speech recognition and natural language processing into its VR devices to allow for better hands free control, which would be particularly useful for non-gaming virtual programs.

The present and future of Facebook AI

Facebook has plenty of lofty goals for the future of its AI project, but the company already uses deep learning and other AI tools in a number of its services. The most visible use of AI on Facebook from a user perspective is the social network’s use of computer vision to suggest which friends should be tagged in a photo.

But Facebook’s visual AI has progressed far beyond simple facial recognition, and it can now caption what is occurring in an image with surprising accuracy. This technology has a wide range of potential uses, and as an example, Schroepfer noted that they could be used to describe images to the visually impaired.

Although Facebook’s AI can often recognize specific objects in an image using computer vision, it has trouble making assumptions that would be relatively easy for humans, for example looking at a pizza and determining whether or not it is vegetarian. Schroepfer explained that for AI to accomplish these sorts of tasks, it will need to develop a sort of “common sense,” which is one of Facebook’s current research goals.

“When computers have common sense they can interact with us in better, more natural ways, from surfacing the most relevant information for us and assisting us with tasks to enabling whole new ways for people to connect,” Schroepfer said. “We’re off to a good start, and I can’t wait to see what tomorrow brings.”

Images courtesy of Facebook