Meta chief artificial intelligence scientist: the generative artificial intelligence technology behind ChatGPT has too many limitations

[chief artificial intelligence scientist of Meta: generative artificial intelligence technology behind ChatGPT has too many limitations] Yann LeCun, chief artificial intelligence scientist of Facebook parent company Meta, pointed out on Thursday that the generative artificial intelligence technology behind ChatGPT has reached a dead end and has too many limitations, and it is too early to talk about their threats. LeCun said: "these systems are full of limitations, they have no understanding of the real world, because they are purely trained by a lot of text content." Most of human knowledge has nothing to do with language. Therefore, artificial intelligence cannot capture this part of human experience. " LeCun added that artificial intelligence systems can now pass the US bar exam, however, artificial intelligence cannot install dishwashers, and a 10-year-old "can learn this skill in 10 minutes".