Facebook’s Pursuit of AI Smarts Leads It to Opening New Research Labs in Pittsburgh and Beyond
Facebook’s artificial intelligence smarts is spreading to Pittsburgh.
The social networking giant said Tuesday it’s opening a new AI research lab in Pittsburgh and has hired several academics specializing in AI who will join the company’s existing offices in Seattle, London, and Menlo Park, CA.
The new hires are part of Facebook’s existing AI research group, which now has about 170 people, said Facebook’s chief scientist of AI Yann LeCun during a press briefing. Facebook, like other tech giants including Google (GOOG) and Microsoft (MSFT), is continuing to incorporate AI technologies like deep learning into its core services and has hired several notable computer science professors over the years in the process.
According to LeCun, Facebook has been setting up its AI research labs near the universities where it has been recruiting, in order to accommodate professors who may not want to move to the company’s home campus in Menlo Park or other remote offices.
“Basically you have to get the talent where it is,” said LeCun. “Not everyone wants to live wherever we have our labs.”
One of Facebook’s new hires, for instance, is Jessica Hodgins, a professor in Carnegie Mellon University’s robotics institute and computer science department. Hodgins, who also previously ran Disney’s research lab out of Pittsburgh, will lead Facebook’s new Pittsburgh AI lab while working at Carnegie Mellon part time.
Although the notion of “dual affiliation” (in which professors split their time between companies and universities) is relatively new to the field of computer science, it is not new to the legal and medical industries, LeCun said. Professors are interested in dual affiliations because they get to retain their academic positions while getting resources like engineering support from companies they wouldn’t get otherwise have access to, he explained.
Dual affiliation also helps downplay the notion that giant tech companies are increasingly poaching the world’s leading AI researchers from universities, thus depriving higher education of experts. Three years ago, for instance, Uber hired about 40 researchers from the National Robotics Engineering Center at Carnegie Mellon, which “was seen as kind of a takeover,” said LeCun.
LeCun said that when recruiting from universities, Facebook works with the school’s administration to ensure that the new hires do not “impede or kill the research” that occurs at the schools. He also explained that Facebook needs these universities to continue teaching students, because those students could eventually become new hires.
“That would be stupid if by establishing ourselves there, we kill the pipeline,” LeCun said. “What’s the point?”
Get Data Sheet, Fortune’s technology newsletter.
As for why Facebook is hiring a robotics expert from Carnegie Mellon, LeCun said that researchers perceive robotics as being one of the most challenging areas for AI.
“There’s pressure in robotics to get machines to learn quickly,” LeCun said. “We don’t have robots that are as agile as a cat or can grab objects. We don’t have robots that can fill and empty your dishwasher.”
The hope is that if researchers are able to create AI that can power more capable robots than today’s versions, Facebook can take the underlying concepts and apply them to other areas of its business or products in unspecified ways. For example, LeCun said that Facebook uses robots to help with data center maintenance.
Additionally, Facebook needs to hire robotics experts because many of the top AI researchers are involved with robots in some way.
“If we don’t work on robotics, we’re basically shutting ourselves off from talented researchers,” LeCun said.