The main line
The kind of artificial intelligence that powers systems like OpenAI’s ChatGPT and Google’s Gemini won’t be able to reach human levels of intelligence, Meta AI chief Yann LeCun told the Financial Times in an interview published on Wednesday, giving insight into as the tech giant plans to develop technology moving forward just weeks after its plans to cash in scared investors and wipe hundreds of billions off its market value.
Meta’s chief AI scientist Yann LeCun said LLMs will not reach human-level intelligence.
Key facts
Large language models, the kind of artificial intelligence that drives popular AI generation tools such as OpenAIs ChatGPT, Metas Llama and Googles Gemini, will never be able to reach human levels of planning and reasoning, the head of Metas AI, Yann LeCun.
The models, often simply called LLMs, are trained using large amounts of data, and their ability to accurately respond to queries is limited by the nature of the data they’re trained on, LeCun said, meaning that they are only correct when given the right. training data types.
LLMs have a limited understanding of logic, don’t have persistent memory, don’t understand the physical world, and can’t plan hierarchically, LeCun explained, adding that they can’t reason in any reasonable definition of the term.
Since they are only accurate when fed the right training data, LeCun, considered one of the three godfathers of AI for his seminal role in the field, said LLMs are also inherently insecure, and that researchers who hoping to develop human-level AI should look to other models. .
LeCun said he and his nearly 500-strong team at the Metas Fundamental Artificial Intelligence Research lab are working to develop an entirely new generation of AI systems based on an approach called world modeling, where the system builds an understanding of the world around him, as people do and develop. a sense of what would happen if something changes based on that.
It could take up to 10 years to achieve human-level AI using the world modeling approach, LeCun predicted.
Receive text alerts of the latest Forbes news: We’ve been sending you text message alerts, so you’ll always know the biggest stories making the day’s headlines. Text to (201) 335-0739 or sign up here.
Key background
Meta, the parent company of some of the world’s most popular tech properties like Facebook, Instagram and WhatsApp, is at the forefront of advances in generative AI. Metas LLMs such as Llama 3 are considered among the most capable in the world, and its Meta AI assistant was recently launched as a standalone app, bringing the company into direct competition with market leaders such as OpenAIs ChatGPT, Anthropics Claude and Googles Gemini . Artificial intelligence models are expensive to train, and investors have responded with less enthusiasm to Meta’s decision to join Silicon Valley’s latest craze. The company lost nearly $200 billion in value in April and triggered a broader decline among tech stocks after CEO Mark Zuckerberg told investors to brace for slower growth as the company pumps funds into AI. He urged investors to be patient as he builds out his AI capabilities at significant expense, despite there being no indication of where, when or even how the investment will pay off. He pointed out that Meta has successfully pursued a similar build and monetization strategy in the past with products like Reels, Stories and News Feed and believes it will benefit from leading the pack with AI as well.
Forbes Rating
Zuckerberg has a fortune of about 165.5 billion dollars. His wealth, much of which comes from his founding stake in Meta’s predecessor Facebook, ranks him as the fourth richest person in the world, sitting behind Bernard Arnault, Jeff Bezos and Elon Musk, with whom he has publicly feuded with and agreed to fight in a cage match. Together with his wife, Priscilla Chan, Zuckerberg has pledged to give away 99% of their shares in Meta within their lifetime.
Further reading
#Chief #Meta #models #ChatGPT #reach #human #intelligence
Image Source : www.forbes.com