[ad_1]
Within a year, his team had developed these models, as well as the tools to design and deploy new ones more quickly. Before, it had taken Quiñonero engineers six to eight weeks to build, train and test a new model. Now it only needed one.
News of the success spread quickly. The team that worked on determining which posts Facebook users would see on their personal news feeds wanted to apply the same techniques. Just as algorithms could be trained to predict who would click which ad, they could also be trained to predict who would like or share which post and then give those posts more importance. If the model determined that a person really liked dogs, for example, posts from friends about dogs would appear higher on that user’s News Feed.
Quiñonero’s success with the News Feed – coupled with impressive new AI research being conducted outside the company – has caught the attention of Zuckerberg and Schroepfer. Facebook now had just over a billion users, making it more than eight times the size of any other social network, but they wanted to know how to continue that growth. The leaders have decided to invest heavily in AI, Internet connectivity and virtual reality.
They created two AI teams. One was FAIR, a basic research lab that would advance advanced technology capabilities. The other, Applied Machine Learning (AML), would integrate these capabilities into Facebook’s products and services. In December 2013, after months of courtesy and persuasion, the executives recruited Yann LeCun, one of the biggest names in the field, to lead FAIR. Three months later Quiñonero was promoted again, this time to lead AML. (It was later renamed FAIAR, pronounced “fire”.)
“That’s how you know what he’s thinking. I was always, for a few years, a few steps from Mark’s office.
Joaquin Quiñonero Candela
In his new role, Quiñonero has built a new model development platform accessible to all Facebook users. Called FBLearner feed, it enabled engineers with little AI experience to train and deploy machine learning models in days. By mid-2016, it was being used by more than a quarter of Facebook’s engineering team and had already been used to train over a million models, including image recognition models, advertising targeting and content moderation.
Zuckerberg’s obsession with getting the whole world to use Facebook had found a powerful new weapon. Teams had previously used design tactics, such as experimenting with content and frequency of notifications, to try to engage users more effectively. Their goal, among other things, was to increase a metric called L6 / 7, the fraction of people who logged into Facebook six of the previous seven days. L6 / 7 is just one of the myriad ways Facebook has measured “engagement” – the propensity of people to use its platform in any way, whether that is by posting things, by posting things to them. commenting, liking or sharing them, or just watching them. . Now, every user interaction once analyzed by engineers was analyzed by algorithms. These algorithms created much faster and more personalized feedback loops to polish and personalize each user’s News Feed to keep increasing engagement numbers.
Zuckerberg, who was seated in the center of Building 20, the main office at Menlo Park headquarters, placed the new FAIR and AML teams at his side. Most of the original AI recruits were so close that their desk and theirs practically touched. It was “the inner sanctum,” says a former executive of the AI ​​organization (the branch of Facebook that contains all of its AI teams), who recalls that the CEO mixed people in and out of his neighborhood in the city. as they gained or lost his favor. “That’s how you know what he’s thinking,” Quiñonero said. “I was always, for a few years, a few steps from Mark’s office.
[ad_2]