- Meta’s chief AI scientist said AI trained on large language models is still not very smart.
- Yan LeCun said AI can’t learn how to load a dishwasher or reason like a child could, CNBC reported.
- AI is not yet as intelligent as animals like dogs or cats, he added.
AI like ChatGPT that’s been trained on large language models isn’t even as smart as dogs or cats, Meta’s chief AI scientist said.
Yann LeCun, an AI “godfather” who received the Turing Award in 2018 for his work in machine learning, made the comments at the Viva Tech conference in Paris this week, CNBC reported.
He said that AI tools trained on large language models are limited because they’re only coached on text. Meta, however, has taken on the challenging task of training AI on video, according to the outlet.
“Those systems are still very limited, they don’t have any understanding of the underlying reality of the real world, because they are purely trained on text, massive amount of text,” LeCun said, per CNBC. “Most of human knowledge has nothing to do with language … so that part of the human experience is not captured by AI.”
Although an AI system can pass the Bar exam for attorneys in the US, it’s still unable to load a dishwasher – something a 10-year-old could “learn in 10 minutes,” LeCun pointed out.
“What it tells you we are missing something really big … to reach not just human level intelligence, but even dog intelligence,” LeCun added. “Until we can do this, we are not going to have human-level intelligence, we are not going to have dog-level or cat-level” intelligence.
LeCun also hit back at experts fanning fears around AI posing a threat to humanity, calling them “preposterously ridiculous,” he told BBC News.
“Will AI take over the world? No, this is a projection of human nature on machines,” he said, adding that computers will definitely become more intelligent than humans one day – but that remained years, if not decades, away.
Read the full article here