(Bloomberg) -- Facebook Inc.’s researchers have created some nifty tools, such as artificial-intelligence software that can recognize objects in digital pictures, or a program that lets people craft three-dimensional images in a virtual environment. Their next step: give the technology away.
“Research and science needs to be done as much as possible in the open,” Chief Technology Officer Michael Schroepfer said. Instead of selling the software, publishing it “makes everyone else move faster. It also helps us validate our work.”
The social network is opening up more of its AI research, seeking to exchange information with other developers so that it can find new insights that could help improve its own products and services. That’s completely opposite from the approach that Apple Inc. has taken; the iPhone maker also works on AI technologies but keeps much of its activity under wraps.
Essentially, Facebook is betting that sharing a technology will help everyone benefit faster than going it alone. Facebook also doesn’t have much of a choice -- it doesn’t control a mobile operating system or a telecommunications network, so it needs to work with other companies to move technology along.
Like it did with data-center hardware, Facebook is opening up its discoveries with the goal of rallying an industry around new technologies. That eventually paid off, helping the company save money on the servers behind the network that connects 1.49 billion users. For artificial intelligence, Facebook’s goal is to nudge other companies in a direction that helps it strategically, Schroepfer said.
The conversations can be mutually beneficial. Facebook has been open about the hurdles it’s facing in artificial intelligence, and semiconductor companies such as Intel Corp. and ARM Holdings Plc have consulted with it on how to make new processors that are optimized for specific AI technologies, such as neural networks.
“They’re coming to us and asking what we need,” Schroepfer said.
Artificial intelligence is becoming an increasingly important technology that’s finding its way into all sorts of endeavors, from language translation and robotic assistants, to health sciences and financial analysis. While the technology has existed for decades, it’s only now that computing and software are becoming advanced enough to begin delivering on some of AI’s lofty goals.
Here are some of the other projects being developed in at Facebook’s Menlo Park, California, headquarters:
* Facebook owns Oculus, the maker of virtual-reality headsets. Using the gadget and hand tools, artists can use specially designed software to build 3-D virtual characters. The company is giving it to design studios. “You don’t have to imagine what this looks like in 3-D -- you are in 3-D creating it,” Scroepfer said.
* Facebook’s new virtual assistant product, M,is learning from the answers that humans give users, with the idea that the AI can eventually take on most of the job.“Without any human intervention, it’s figured out that if you ask us, ‘I’d like to send flowers,’ the two most important questions we need to respond with are, ‘What’s your budget?’ and ‘Where would you like them sent?”’ he said. Eventually the AI will become reliable enough to work with all of Facebook’s 1.49 billion users. “The reason this is exciting is it’s scalable.”
* Facebook AI software is teaching computers how to look at pictures and determine what items should be thought of as separate objects, so it knows, for example, that a coffee cup isn’t physically connected to the table it’s sitting on. Facebook plans to present a research paper on image segmentation at the influential Neural Information Processing Systems conference in December. “Once you can get a better mapping of what’s in an image, you can get to tons of useful tools” for news feed, Schroepfer said. People would be able to personalize their feeds by type of image. “I might want to tell Facebook, ‘love the baby photos, love the sports, tune down the latte art.”’
Facebook’s open approach also helps the company quickly apply technological breakthroughs to working products, which makes it easier to recruit good engineers, Schroepfer said.
“The promise I made to all the AI folks that joined us is we’re gonna be the best place to get your work to a billion people as fast as possible,” he said.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access