AI forms backbone of Facebook’s 10 year plan

Facebook founder Mark Zuckerberg

Facebook founder Mark Zuckerberg

Facebook has seemingly positioned artificial intelligence as one of the catalysts for innovation for the company over the next 10 years.

Outlining its technology roadmap for the next 10 years, the company highlighted artificial intelligence, as well as virtual and augmented reality, as technologies to drive new features and user experience. New features highlighted include translation, photo image searches, ‘talking pictures’ and real-time video classification.

“Artificial Intelligence will power all kinds of different services with better than human level perception and we’ll see the emergence of the next major computing platform in virtual and augmented reality,” said Facebook founder Mark Zuckerberg during his opening keynote at F8. “These are all elements of our 10 year roadmap to connect the world and each of these elements is in service of our mission. It’s about bringing people together, that’s what we do here.”

The company has been making efforts in recent months to bolster its position in the artificial intelligence space. Recently, the company announced a string of new hires for its artificial intelligence research team, including a number of acquisitions from Microsoft’s R&D team, another company who have been making strides to perfect AI. The new staff members bring experience to the team in the fields of causal inference in learning systems, computer vision, cost-sensitive learning, speech recognition and syntactic parsing with approximate inference.

Joaquin Quiñonero Candela, Director of Applied Machine Learning at Facebook

Joaquin Quiñonero Candela, Director of Applied Machine Learning at Facebook

On the company blog, Joaquin Quiñonero Candela, Director of Applied Machine Learning at Facebook outlined a number of use cases which are a reality today. “We built an AI backbone that powers much of the Facebook experience and is used actively by more than 25 percent of all engineers across the company. Powered by a massive 40 PFLOPS GPU cluster that teams are using to train really large models with billions of parameters on huge data sets of trillions of examples, teams across the company ae running 50x more AI experiments per day than a year ago, which means that research is going into production faster than ever.”

The company has already applied an AI-based automatic translation system, claiming that an off-the-shelf translation program would not be adequate as they trained on a general corpus like appliance manuals. As language on Facebook is far more colloquial, the systems would not be effective. The company claims that its AI capabilities have the ability to learn and recognize new expressions, regional differences and the various uses of emojis.

In terms of pictures, Facebook claim its AI can understand the content of the image at a pixel level to make classification and searching of image simpler. “This is called image segmentation, and it allows us to recognize individual objects in the image as well as their relation,” said Candela. “Using image segmentation we will be able to build more immersive experiences for the visually impaired with “talking images” you can read with your fingertips, as well as more powerful ways to search images. In one case here, we have the ability to search for ‘a photo of us five on skis on the snow, with a lake in the background and trees on both sides’.

“AI is central to today’s Facebook experience, and, with our research pushing the state of the art, we’re just getting started on this journey. I’m excited to see where it takes us next.”