'Complete B.S.': Meta's Yann LeCun dismisses fears of AI threat
What's the story
Yann LeCun, a well-known AI expert and senior researcher at Meta, has dismissed fears of artificial intelligence (AI) being an existential threat to humanity.
Speaking to The Wall Street Journal, LeCun said the notion of AI becoming intelligent enough to threaten humans is "complete B.S."
The statement is in line with his previous expressions of skepticism on this issue.
AI shortcomings
LeCun highlights AI's limitations in comparison to house cat
LeCun, also a New York University professor and Turing Award laureate, has previously voiced his skepticism about the advancement of AI.
He once tweeted that before worrying about controlling super-intelligent AI, "we need to have the beginning of a hint of a design for a system smarter than a house cat."
This comparison highlights his belief in the current limitations of AI systems.
Deficiencies
LLMs criticized for lack of key capabilities
In the interview, LeCun slammed today's large language models (LLMs) for not being able to even match a cat's intelligence.
He noted that these models don't have basic capabilities such as persistent memory, reasoning, planning, and understanding the physical world.
LLMs, according to him, only prove that "you can manipulate language and not be smart," debunking the notion that they could ever lead to true artificial general intelligence (AGI).
Future prospects
LeCun suggests new approaches for AGI development
Despite his criticism of current AI models, LeCun isn't completely dismissive of the possibility of AGI.
He indicated that new strategies will be needed to reach this level of AI sophistication.
For instance, he pointed out the work being done by his Fundamental AI Research team at Meta on processing real-world video data.