Facebook says it found faster way to translate thru AI
News Feed chief explains new ‘rocket ship’ feature
NEW YORK, May 10, (Agencies): Facebook says its researchers have found a new way to use artificial intelligence to translate material on its social network faster and more accurately.
This could mean Facebook users eventually seeing everything translated immediately into their preferred language, not just post but videos too. Facebook already translates posts in more than 45 languages, but CEO Mark Zuckerberg says there is still “a lot more to do.”
For now, Facebook has made the research and its methods publicly available so developers and others can use it to build translation and other language tools. Beyond language translation, the technology can be used for chatbots, for example, or other language-based tasks.
The method uses something called a convolutional neural network, a technology that’s already used for image processing and other types of machine learning.
Meanwhile, some Facebook users around the world have recently noticed a rocket ship icon that has popped up next to the News Feed tab on the menu bar. Facebook executives have kept mum about the new feature, until now.
Adam Mosseri, Facebook’s head of News Feed, described the rocket ship icon as an experiment with a relatively small number of users to expose them to information from people and pages that they have not chosen to like or follow. Mosseri, who has been at Facebook for nearly a decade, spoke Monday night in conversation with Variety co-editor-in-chief Andrew Wallenstein during the publication’s kickoff dinner leading in to its Entertainment and Technology Summit on Tuesday in New York.
“People can learn about new stories from sources they are not connected to,” Mosseri said of the tab, which he characterized as in keeping with the overall mission of News Feed: “to connect people with the stories that matter to them.” The information stream in the rocket ship tab is curated by Facebook based on algorithms that deduce a user’s interests by their Facebook usage patterns.
Mosseri explained that Facebook’s recommendation architecture is based on “educated guesses” that come from the tidal wave of user data that is constantly studied. No detail is too small to ignore, he said. “We look at little things, like when you watch a video, do you change it to full screen, do you turn up the volume?” he said. “We try to look for patterns.”
Mosseri said the space vessel motif was chosen by a designer on the project who Mosseri jokingly described as “very, very Irish.” At first executives figured it was a placeholder image.