Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

Meta's top AI scientist debunks overhyped AI predictions

Dec 04, 2023 - businessinsider.com
Meta's chief AI scientist, Yann LeCun, has expressed his skepticism about AI reaching human-level intelligence in the near future, stating that it will likely only achieve "dog-level" or "cat-level" intelligence. He countered claims made by Nvidia's CEO, Jensen Huang, who recently predicted that artificial general intelligence will be achieved in the next five years. LeCun argued that AI is being trained on language models and text that are insufficient for creating advanced AI, and that even with extensive training, AI systems still lack basic understanding of the world.

LeCun has previously stated that AI's capabilities are currently "limited" and that it is not even as smart as cats or dogs. He noted that while AI can pass a lawyer's bar exam due to its language training, it cannot perform simple physical tasks like unloading a dishwasher. He also dismissed fears about AI taking over humanity as "preposterously ridiculous" and overstated.

Key takeaways:

  • Meta's chief AI scientist Yann LeCun is sceptical that AI will reach human-level intelligence anytime soon, predicting it will achieve "dog-level" or "cat-level" intelligence first.
  • LeCun countered claims made by Nvidia's CEO Jensen Huang, who recently said that artificial general intelligence will likely be achieved in the next five years.
  • LeCun believes that AI is being trained on language models and text that are not sufficient to create advanced AI, and that most of human knowledge is not captured by AI.
  • LeCun has also dismissed fears that AI could take over humanity, calling such predictions "preposterously ridiculous" and overstated.
View Full Article

Comments (0)

Be the first to comment!