Yann LeCun AI skepticism

Yann LeCun, chief AI scientist at Meta and considered “one of the godfathers” of artificial intelligence, offers a controversial perspective in the industry. Yann LeCun’s AI skepticism highlights his critical approach to evaluating the progress of artificial intelligence. In an exclusive interview with The Wall Street Journal, LeCun shares his refreshing skepticism regarding the potential of AI.

“Today’s AI Models Are Far from Rivaling Even Cat Intelligence”

“Today’s AI models are far from rivaling even the intelligence of a cat, let alone ours,” says Yann LeCun, emphasizing the limitations of current AI technology. As chief AI scientist at Meta, LeCun oversees one of the best-funded AI research organizations in the world. In an interview with The Wall Street Journal, he provides an unfiltered view of AI’s current abilities and shortcomings.

LeCun was born and raised in Paris. His interest in AI began with HAL 9000, the malevolent AI from Stanley Kubrick’s movie “2001: A Space Odyssey.”” Learn more about artificial intelligence. LeCun earned his Ph.D. at the Sorbonne and worked at Bell Labs before joining NYU as a computer science professor. His contributions to AI research have earned him numerous accolades, such as the A.M. Turing Award.

A Voice of Reason in the AI Debate

When asked whether we should fear AI surpassing human intelligence, LeCun responds bluntly: “Pardon my language, but that’s complete nonsense.” His views challenge the idea that AI will soon outmatch human capabilities. He has publicly debated business leaders and fellow scientists, including Elon Musk and Geoffrey Hinton, about the potential dangers of AI. Hinton, a friend of almost 40 years, recently received the Nobel Prize in Physics and has warned of AI’s existential threats. LeCun disagrees, saying: “Today’s AI models are far from rivaling the intelligence of our pets, let alone ours.”

Current AI Systems Lack Fundamental Qualities

LeCun argues that current AI systems are missing fundamental qualities such as reasoning, planning, and common sense. Moreover, these are key attributes required to advance AI to the next level. “We need the first hints of a design for an AI system that is smarter than a household cat,” he says. According to LeCun, cats have a mental model of the physical world, long-term memory, and a certain capacity for reasoning and planning—none of which are present in today’s so-called “advanced” artificial intelligences, including those created by Meta itself.

The Misguided Pursuit of Artificial General Intelligence (AGI)

LeCun dismisses the notion that simply scaling up AI models will lead to artificial general intelligence (AGI). He believes the problem with today’s AI systems lies in their design, not their size. “No matter how many GPUs tech giants cram into data centers around the world,” he says, “today’s artificial intelligences won’t bring us to AGI.”

AI Systems Inspired by Human and Animal Learning

LeCun advocates for researching fundamentally different AI architectures, drawing inspiration from how animals learn and perceive the world. For an in-depth analysis, visit our AI research innovations page. For more information, visit our AI research page. He is particularly excited about projects in Meta’s FAIR lab that process real-world video to create AI systems more similar to humans. LeCun envisions future AI systems that will possess human-level traits, such as common sense and the ability to behave like human assistants. Additionally, these systems must demonstrate adaptive behavior to effectively support users. However, he believes achieving this will take decades and require a more nuanced approach than merely investing resources in large language models.

“In the future, when people interact with AI systems, smart glasses, or other devices, these AI systems will need to exhibit human-level qualities. Therefore, they must possess common sense and effectively support users in various scenarios. They must have common sense and behave like a human assistant,” he explains.