- Facebook AI Research boss Yann LeCun said that the latest AI breakthroughs should not be over-hyped.
- He said self-driving cars and game-playing AI agents are examples of “narrow AI.”
- We’re not even close to the “general AI” machines depicted in Hollywood movies, he said.
There’s no question that machines are getting smarter every year but we shouldn’t overestimate their abilities just yet.
That was the message of Yann LeCun, the head of Facebook AI Research (FAIR), in an interview with The Verge published on Thursday.
While machines can learn some things for themselves and beat humans at board games like Go (which has more possible moves than there are atoms in the universe), they’re still nowhere near as intelligent as a baby, or an animal, according to LeCun.
“We’re very far from having machines that can learn the most basic things about the world in the way humans and animals can do,” LeCun reportedly told The Verge.
“Like, yes, in particular areas machines have superhuman performance, but in terms of general intelligence we’re not even close to a rat.”
Artificial general intelligence — the ultimate goal for many AI researchers — refers to computer systems that posses intelligence comparable to that of the human brain.
LeCun highlighted that all of the recent AI breakthroughs relating to things like self-driving cars and interpreting medical images are examples of “narrow AI,” not “general AI.”
He said: “So for example, and I don’t want to minimise at all the engineering and research work done on AlphaGo by our friends at DeepMind, but when [people interpret the development of AlphaGo] as significant process towards general intelligence, it’s wrong. It just isn’t.”
LeCun also warned journalists not to mislead the public by using Terminator photos in their stories or over-hyping breakthroughs in the field. “I keep repeating this whenever I talk to the public: we’re very far from building truly intelligent machines.”