AI pioneer Yann LeCun warns business tools built on LLMs are headed for obsolescence

This AI Godfather Says Business Tools Built on LLMs Are Doomed

And when that happens, LeCun says, even more investment will be required to create the superintelligence technology that will replace LLM-based AI -- systems that he says should already be the focus of development efforts and funding. While that may seem like an outlier view, LeCun, 65, is far from a tech outsider. The Turing Award winner ran Meta's AI research unit for a decade, only leaving last November to launch his own Paris-based startup, Advanced Machine Intelligence Labs (AMI Labs). In addition to disliking the managerial duties that came with the research-rooted Meta job, LeCun said his departure was motivated by his view that Silicon Valley has prioritized short-term business interests over far more important and attainable scientific objectives. Top of those commercial concerns he cites was developing and marketing LLM-based AI chatbots and apps with limited capabilities, rather than superintelligence systems with virtually boundless potential.

An AI pioneer warns the tech 'herd' is marching into a dead end

AI pioneer Yann LeCun argues large language models are a dead end for achieving human-level intelligence. After leaving Meta, he criticizes Silicon Valley's herd mentality, defends open source and pursues alternative AI systems that can plan, learn from the real world and rival global competitors. Throughout his 40-year career as a computer scientist, Yann LeCun has earned a reputation as one of the world's leading experts on artificial intelligence and a man with a penchant for throwing verbal grenades. He was one of three pioneering researchers who received the Turing Award, often called "the Nobel Prize of computing," for their work on the technology that is now the foundation for modern AI. For more than a decade, he also served as chief AI scientist at Meta, the parent company of Facebook and Instagram. But after leaving Meta in November, LeCun has become increasingly vocal in his criticism of Silicon Valley's single-minded approach to building intelligent machines. He argues that the technology industry will eventually hit a dead end in its AI development -- after years of work and hundreds of billions of dollars spent. The reason, he said, goes back to what he has argued for years: Large language models, the AI technology at the heart of popular products like ChatGPT, can get only so powerful. And companies are throwing everything they have at projects that won't get them to their goal to make computers as smart as or even smarter than humans. More creative Chinese companies, he added, could get there first. "There is this herd effect where everyone in Silicon Valley has to work on the same thing," he said during a recent interview from his home in Paris. "It does not leave much room for other approaches that may be much more promising in the long term." That critique is the latest shot in a debate that has roiled the tech industry since OpenAI sparked the AI boom in 2022 with the release of ChatGPT: Is it possible to create so-called artificial general intelligence or even more powerful superintelligence? And can companies get there using their current technology and concepts? Few scientists have as much history with the topic as LeCun, 65. Much of what the tech industry is trying to do now has its roots in an idea that he has nurtured since the 1970s. As a young engineering student in Paris, he embraced a concept called neural networks, even though most researchers thought the idea was hopeless. Neural networks are mathematical systems that learn skills by analyzing data. At the time, they had no practical use. But a decade later, when he was a researcher at Bell Labs, LeCun and his colleagues showed that these systems could learn to read handwriting scribbled on envelopes or personal checks. By the early 2010s, researchers had begun to show that neural networks could power a wide range of technologies, including face recognition systems, digital assistants and self-driving cars. As Google, Microsoft and other tech giants placed big bets on the idea, Facebook hired LeCun to build an AI research lab. Not long after ChatGPT was released, the two researchers who received the 2018 Turing Award with LeCun warned that AI was growing too powerful. Those scientists even warned that the technology could threaten the future of humanity. LeCun argued that was absurd. "There was a lot of noise around the idea that AI systems were intrinsically dangerous and that putting them in the hands of everyone was a mistake," he said. "But I have never believed in this." LeCun also helped push Meta and its rivals to freely share their research through academic-style papers and so-called open source technologies. As more people said AI could be a threat of some sort to humans, a number of companies curtailed their open source efforts. But Meta kept going. LeCun repeatedly argued that open source was the safest path. It meant that no one company would control the technology and that anyone could use these systems to identify and fight potential risks. Now as a number of companies, including Meta, appear to be moving away from that method because they want an edge over their rivals and continue to worry about dangerous uses, LeCun is warning that American companies could lose their lead to Chinese rivals that are still using open source. "This is a disaster," he said. "If everyone is open, the field as a whole progresses faster." Meta's AI work ran into a snag last year. After outside researchers criticized the company's latest technology, Llama 4, and accused Meta of misrepresenting the power of the system, Mark Zuckerberg, Meta's CEO, spent billions on a new research lab dedicated to the pursuit of "superintelligence" -- a hypothetical AI system that exceeds the powers of the human brain. Six months after the creation of the new lab, LeCun left Meta to build his own startup, Advanced Machine Intelligence Labs, or AMI Labs. Even though his research laid the groundwork for LLMs, LeCun argued that they were not the final answer to AI development. The problem with current systems, he said, is that they do not plan ahead. Trained solely on digital data, they do not have a way of understanding difficulties in the real world. "LLMs are not a path to superintelligence or even human-level intelligence. I have said that from the beginning," he said. "The entire industry has been LLM-pilled." During his last several years at Meta, LeCun worked on technology that tried to predict the outcome of its actions. That, he said, would allow AI to progress beyond the status quo. His new startup will continue that work. "This type of system can plan what it is going to do," he said. "Current systems -- LLMs -- absolutely cannot do that." Part of LeCun's argument is that today's AI systems make too many mistakes. As they tackle more complex tasks, he argued, mistakes pile up like cars after a collision on a highway. But over the past several years, these systems have steadily improved. And in recent months, the latest models, which are designed to "reason" through questions, have continued to improve in areas like math, science and computer programming. "These models do make mistakes. But we have shown that a system can try many different options -- in its own head, so to speak -- before settling on a final answer," said Rayan Krishnan, the CEO of Vals AI, a company that tracks the performance of the latest AI technologies. "Progress is not slowing down. It has become clear that language models can take on new tasks and continue to get better at doing everything we want them to do," he said. Subbarao Kambhampati, an Arizona State University professor who has been an AI researcher nearly as long as LeCun, agreed that today's technologies don't provide a path to true intelligence. But he also pointed out that they were increasingly useful in highly lucrative areas like computer coding. LeCun's newer methods, he added, are unproven. For LeCun, that is why his new company is important. The past several decades, he said, are filled with AI projects that seemed like a way forward but ran out of steam. And Silicon Valley is not guaranteed to be the winner in this global race. "The good ideas are coming from China," he said. "But Silicon Valley also has a superiority complex, so it can't imagine that good ideas can come from other places."