Facebook's chief AI Scientist tells why we're currently experiencing a key moment for the future of this technology
- According to Yann LeCun, artificial intelligence is not yet advanced enough to design a machine capable of acquiring common sense.
- “A rat has more common sense than the best artificial intelligence system we can build,” says the man who has just been promoted to Facebook’s chief ai officer.
- There is a risk that the people who fund AI research will lose interest in it if no progress is made in this direction “in the next two or three years,” according to him.
Artificial intelligence has recently made many advances in translation, speech recognition or in the development of the autonomous car.
Advances that fuel fantasies about how AI will shape the world of tomorrow. For example, Elon Musk tweeted in September 2017 that it would be “the most likely cause of a third world war”. As for the much-hyped Sophia robot, it suggests that we have entered the era of humanoid robots, as Le Monde writes.
Yann LeCun, newly appointed scientific director of Facebook’s AI, seeks to “moderate the hopes” raised:
“A rat has more common sense than the best artificial intelligence system we can build”, he said at a conference on the sidelines of the Connexions exhibition organized by Facebook on Tuesday, January 23 in Paris.
That is, the breadth of knowledge in AI does not yet allow to design a machine that can acquire common sense, have a “conscience”.
“We’re not going to have this unless we figure out how to make machines learn how the world works by observing it,” he explained in an interview with The Verge in October 2017. “You know, just watching videos and reading books. And that’s the scientific and technological challenge for the next few years. I call this predictive learning.”
Such an advance is a holy grail for experts and researchers because it conditions the future of AI. Indeed, it would prevent a new “winter” – those periods during which there is no major breakthrough in the field because of the lack of research funding.
During its history, which begins in the 50s, AI has experienced several “winters”, as Les Echos recalls.
“If we can make significant progress in the next two or three years, I think we won’t have an AI winter problem,” Yann LeCun said at the conference.
“If it takes 5, 10 or 15 years, it’s possible that all these people who are funding all this research will start to lose patience and wonder if it’s really worth it. The question is whether people will believe in it until we find something that is meaningful progress.”
Recently, Yann LeCun has been attacking the infamous Sophia Robot (and Business Insider) for spreading what he considers “absolute bullshit” about artificial intelligence.
Mark Zuckerberg has made technology one of the cornerstones of his company’s evolution.
Receive our latest news
Every day, the main Business Insider news