The ChatGPT conversational robot is “not at all revolutionary” and the big progress will come rather from automatic personal assistants, believes a French Facebook executive.
Asked on France Inter on April 12 about the application developed by OpenAI, the director of the Meta group’s artificial intelligence laboratory, Yann Le Cun, replied: “Yes, it’s a good product. science and technology behind it, it’s not revolutionary at all, no”.
In the “absolutely huge amounts of text” that caused ChatGPT to answer questions from its users, “there is a lot, a lot of human knowledge, but it is very superficial”, explained this researcher. “It doesn’t include, for example, knowledge of the physical world, the real world. So these systems, in a way, have a lot less knowledge of the real world than your cat does,” he said.
Augmented reality glasses and virtual assistants
Asked about the most important innovations foreseeable for the next 10 to 15 years, this New York University professor mentioned two types of products. Yann Le Cun believes in the arrival of “augmented reality glasses, which will therefore be able to superimpose virtual images on the real world”, and for example transcribe live in our language a conversation in an unknown foreign language.
In addition, according to him, “we will interact with intelligent virtual assistants who, at certain times, will have intelligence similar to human intelligence, perhaps superior in certain areas, and who can help us in our lives in all days”, for example by allowing us “to focus on things that interest us, and not having to spend an hour on the phone to talk to your plumber or your Social Security administration”.
This leader of Meta finally said he was hostile to the petition signed in March by personalities such as entrepreneur Elon Musk or historian Yuval Noah Harari to “take a break of at least six months in the training of computer systems. ‘AI more powerful than GPT-4’. “It’s a kind of new obscurantism. I don’t see any good motivation at all to try to limit progress (…) It’s completely unrealistic to tell people: stop research for six months. way, no one is going to do it,” retorted Mr. Le Cun.
