Facebook Inc. has announced plans to expand its artificial intelligence expertise with the establishment of a research team in Paris.
According to the company, the new center will build upon the work already carried out at Facebook AI Research (FAIR) centers in Menlo Park, Calif., and New York by focusing on “ambitious long-term research projects” across a diverse range of subjects, including “image recognition, natural language processing, speech recognition, and the kinds of physical and logical infrastructure required to run these AI systems.”
The social networking giant also stated that its hopes its research will “ultimately help us to make services like News Feed, photos, and search even better and enable an entirely new set of ways to connect and share.”
In recognition of the fact that France contains some of the best AI researchers in the world, Facebook plans to “work openly with and invest in the AI research community in France, the EU, and beyond” as part of a concerted effort to make “meaningful progress” in these subject areas.
French AI researcher Yann LeCun
“We have a collaborative agreement in place with INRIA, a leading research institute in Paris, which opens up new joint study opportunities for talented research professionals, Ph.D. students, and postdoctoral researchers. We believe this open model ultimately spurs more innovation, encourages collaboration and mutual review, and helps us all move faster,” added the company.
‘Scary products’
According to AI expert Allan Tucker, senior lecturer in the Department of Computer Science at Brunel University in London, the Paris research center could be a valuable addition to the burgeoning AI sector in France and help the company to form a bridgehead in Europe to keep up with rival Google Inc.
“I think this is clearly a push to compete with Google in London,” he said. “Their acquisition of DeepMind at Kings Cross, as well as the new HQ planned nearby, is clearly going to create a center of excellence, and Facebook will want to have a similar competing presence within Europe.”
“AI tools are going to be key in the development of new algorithms for social media, [via] image analysis, pattern recognition, as well as other technological developments including drones and the Internet of Things,” Tucker added.
For instance, Facebook’s Moments app allows users to create shared photo albums using AI for facial recognition. The app was built by its Creative Labs team.
Unlike Google’s hallucinogenic images, another Facebook AI project was able to generate images that looked realistic to human viewers.
Facebook also hopes that new technology will help it “eliminate spam and, eventually, violent videos,” said LeCun. European regulators are examining Facebook’s potential privacy violations, but AI could open up new advertising opportunities.
For Tucker, the creation of what he described as “more responsive industries online” is the “immediate win” of such initiatives for companies like Facebook, Google, and Baidu chiefly through the acquisition of business intelligence and better marketing. AI could also lead to new technologies, which he argued will offer users and customers more “interesting, cool and probably scary products.”
“I still think robotics development being profitable is further away, though drones may prove me wrong,” Tucker said. “Autonomous vehicles are being pushed in the media, but I am still sceptical — litigation may still hold this all back. Decision-support tools in medicine, engineering, and other areas of human expertise are already pushing ahead, and centers for excellence like London and Paris can only improve these.”
More generally speaking, Tucker also believes that all AI-based robotics technologies hold promise for commercial growth. “The real advances [lie in] using all of these techniques on the data being collected and shared online,” he said.
“The technologies are improving year on year, but the step-change in my opinion is in the way data is collected,” Tucker explained. “Mobile device advances have driven an improvement in the way existing AI technologies can be used. Deep learning has really been around a long time but can now be used in areas that were not possible due to limited data.”
More on AI, Robotics, and Facebook:
Facebook M can place orders for users
Facebook’s new M is an extension of the Facebook Messenger app and uses AI and humans as a virtual assistant. The service, which is in testing, is intended to expand the company’s reach beyond social media.
“Unlike other AI-based services in the market, M can actually complete tasks on your behalf,” said David Marcus, vice president of messaging products at Facebook. “It can purchase items, get gifts delivered to your loved ones, book restaurants, travel arrangements, appointments and way more.?
Facebook M, which will initially rely on human “trainers” for any questions the AI can’t answer, is intended to compete with Apple’s Siri, Microsoft’s Cortana, and Google Now.
