French AI pioneer Yann LeCun is not a fan of ChatGPT

French AI pioneer Yann LeCun is not a fan of ChatGPT

Meta’s “Chief Artificial Intelligence Scientist,” winner of the prestigious Turing Award for his pioneering work in deep learning, wanted to define an “I” about ChatGPT and its designer, OpenAI.

There is nothing revolutionary about it, although this is what the public perceives During an online conference that ZDnet was able to attend, Yann LeCun wanted to temper the fervor that has animated the Web since ChatGPT launched at the end of November.

His speech is interesting, because Yann LeCun is far from anyone. The work of the French researcher, who is now the chief artificial intelligence scientist at Meta, is indeed the origin of the technology Deep learning, or deep learning, which has revolutionized artificial intelligence since the end of the 2000s. Search it out It even earned him the prestigious Turing AwardNobel Prize in Computing.

Read also: Five tips for getting better answers with ChatGPT

“Well put together, but not particularly innovative”

ChatGPT is also based on a file Deep learning for work. But it does not represent, according to LeCun, a major technical advance. ” When it comes to the underlying technologies, ChatGPT isn’t particularly innovative “Especially mentioned during the conference, recognizing that ChatGPT” Well, well done “.

As LeCun also reminds us, OpenAI is actually far from the only one working on AI systems of this kind, and it’s not necessarily ahead of other research labs. Meta, his employer, also designed a file “Awesome language model”, which responds to the sweet name OPT-175B. It is comparable in size to GPT-3, the model behind ChatGPT, and has been freely given to the scientific community. Google is also working on similar technologies, and they’ve already made it Public offering of the LaMDA bot nearly two years ago.

READ  Ubisoft + on Xbox: The Service May Arrive Early Next Week! | Xbox One

“It’s not just Google and Meta: Six startups have fundamentally very similar technology LeCun said, according to ZDnet. Before we remember that the technology behind the GPT-3 model comes from the work on self-supervised learning, which is the technology introduced by LeCun Advocacy has been around for a long time, even before OpenAI existed “… but also ” transformers “Technique Deep learning Introduced by Google researchers in 2017. Not to mention the use of a reactions Human (RLHF) Opened by DeepMind – Currently owned by Google – In 2017.

“ChatGPT and other great language paradigms didn’t come out of nowhere, they are the result of decades of contributions from various people” LeCun sums up on Twitter.

But what are Google and Meta doing?

However, one question remains: If this technology is ready, especially in companies like Google or Meta, why do they leave it in the closet? According to LeCun, the answer is quite simple: OpenAI is a startup whose economic equation is very different from that of more established groups. When asked about this by the user, He replies on Twitter :” Large companies could have made a public demonstration, but they did not, because they have less profit than a small company looking for investors, and a lot to lose (due to bad press). »

READ  WhatsApp now lets you leave a group secretly

And it’s true that when it comes to bad press, Meta really knows a lot. However, Mark Zuckerberg’s company has launched an experiment with BlenderBot, but this bot is only available in the US and was criticized upon its launch. to embrace conspiracy theories. As for Google, it is clearly afraid of unleashing a monster that would damage its reputation … and his wallet.

This should change soon: LeCun does not hesitate to say that Meta will double its generative AI services in the future. And at Google, it is commotion To quickly counter OpenAI initiatives. ChatGPT may not be revolutionary, but it shook things up in a matter of weeks.

Source :

Zdnet.com

Leave a Reply

Your email address will not be published. Required fields are marked *