Meta AI head Yann LeCun discussed the state of AI and quantum computing across the industry during a recent event celebrating the 10th anniversary of the founding of Meta’s Fundamental Artificial Intelligence Research (FAIR) team. .
During LeCun’s commentary, he commented on Nvidia’s current dominance over the AI hardware industry, the likelihood of human-level AI emerging in the near future, and why Meta is not currently pursuing quantum computing alongside its competitors.
By “not in the short term” I mean “clearly not in the next 5 years”, contrary to what many people in the AI industry say.
Yes, I am skeptical of quantum computing, particularly as it applies to AI. https://t.co/5t63w1GNfL– Yann LeCun (@ylectun) December 3, 2023
The war of artificial intelligence
LeCun’s views on the imminence of so-called human-level AI are well documented. He has long maintained that we will need to achieve “dog” and “cat” level AI before the field advances enough to support human level AI. And so far he has been reluctant to give predictions about when those first milestones will occur.
By comparison, Elon Musk recently made the bold prediction that a “digital God” would arrive in the next 3 to 5 years.
In the middle, perhaps, is Nvidia CEO Jensen Huang. He recently claimed that AI could complete tests “fairly competitive” with humans in the next five years. While this stops short of claiming full human-level capabilities, some experts view testing as a measure of an AI’s capabilities.
LeCun, apparently, is not among them. During FAIR’s 10th anniversary celebration, he provided comments related to the Nvidia CEO’s claims:
“I know Jensen. There is an AI war and he is supplying the weapons.”
The Meta AI boss’ statement likely refers to the recent news that Nvidia is now the world’s most valuable chipmaker, a feat largely attributed to the company’s GPUs becoming the status quo hardware for training. great language models like ChatGPT.
LeCun further explained that the current technology cited as a potential spark for AGI, generative AI, was simply not good enough. “Text is a very poor source of information,” he said, adding, “train a system with the equivalent of 20,000 years of reading material, and it still doesn’t understand that if A is the same as B, then B is the same.” as a.”
Related: Nvidia posts record revenue of $18 billion in Q3, cites generative AI as key driver
Quantum computing
Another area that LeCun addressed during the event was quantum computing. Unlike its competitors Google and Microsoft, Meta has relatively distanced itself from the quantum computing race.
“Quantum computing is a fascinating scientific topic,” LeCun said, but other comments made clear that, in his view, the technology was not mature. “The number of problems that can be solved with quantum computing can be solved much more efficiently with classical computers.”
Meanwhile, Microsoft recently signed a $100 million partnership with Canadian quantum computing company Photonic to bring a fully functional, fault-tolerant quantum networking system to market within the next five years.
