The Blubber

Elon Musk has said:

AI will probably be smarter than any single human next year. By 2029, AI is probably smarter than all humans combined.

LeCun, Meta's chief AI scientist and often regarded as one of the "godfathers of AI," rebutted Musk's argument and took a swipe at Tesla's self-driving cars in development.

"No," he stated on X. "If that were the case, we would have AI systems that could learn to drive a car in 20 hours, like any 17-year-old."

"Yet, we still lack fully autonomous, reliable self-driving cars, despite having millions of hours of labeled training data," LeCun added.

The French-American scientist noted that AI models possess roughly the same computing power as a common housecat's brain but are not nearly as intelligent.

"Why aren't those systems as smart as a cat?" LeCun pondered. "A cat can remember, understand the physical world, plan complex actions, and engage in some level of reasoning—actually, much better than the largest LLMs," he remarked.

For LeCun, the challenge goes beyond simply scaling up existing technology; there's a fundamental missing element preventing machine learning and language models from reaching their full potential.

"This indicates that we are lacking something conceptually significant to achieve the intelligence of animals and humans," LeCun explained.

"At some point in the future, these systems might actually be smart enough to provide you with useful information better than what you can obtain from a search engine. But that's not the case today."

In contrast, Musk has voiced concerns about AI's potential dangers, suggesting that it could develop super-intelligence capable of surpassing human capabilities.

"While the likelihood of humanity's annihilation is small, it's not zero," the billionaire previously told The Wall Street Journal.

"There is a risk that advanced AI could either eliminate or restrict humanity's progress," he has said.