Yann LeCun, former chief AI scientist at Meta, helped lay the foundation for the modern AI boom long before chatbots became an industry craze. Now, next Leaving dead And in late 2025, he will build a multibillion-dollar startup on the premise that the race to super-AI starts in the wrong place. Speaking at the VivaTech conference on June 17, LeCun Share more details On why he believes his startup, Advanced Machine Intelligence (AMI Labs), is on a better path to human-level AI. Earlier this year, the Paris-based company It raised $1.03 billion At a valuation of $3.5 billion, giving LeCun and his team new support to pursue an alternative to current generative AI models developed by tech giants.
“I think there is a need for new models to go beyond the limitations of current systems,” LeCun, 65, told Wired’s Steven Levy on stage.
Building better language models, as many AI companies are doing, could be one path, but “it’s kind of a slow path,” LeCun said. According to LeCun, AMI Labs is building “universal models” that learn from reality, understand consequences, predict what will happen next, and choose the best actions based on those predictions. “We have high hopes that this will be a complete change in the design of smart systems,” Lacon said.
LeCun believes that most companies in Silicon Valley have become “full of LLMs” — or overly convinced that scaling up big language models will eventually produce human-like intelligence. The result, he said, is a “monoculture” of AI systems that is powerful in some areas but remains fundamentally limited.
Although LLM degrees do have strong capabilities in areas such as programming, they are language-based. Real-world data from cameras, robots and other sensors is messier and harder to predict, which is why he believes a different approach is necessary. For AMI labs, the result may be: “There are vastly more intelligent systems that can predict the consequences of their actions, that can plan, that can think, and may even reach human intelligence,” he said.
Coming from LeCun, the criticism carries weight. A longtime New York University professor, he is among the three researchers often referred to as the “Godfathers of AI.” The others are Geoffrey Hinton, a University of Toronto professor and former Google employee, and Yoshua Bengio, a University of Montreal professor who founded the Milla Institute for Artificial Intelligence in Quebec. In 2018, they did He won collectively The prestigious Turing Award for achievements that helped push deep learning into the mainstream.
From FAIR to AMI Laboratories
In 2013, to be Join Facebook To launch what would become Facebook AI Research, or FAIR, he later moved on to become Chief AI Scientist at Meta. He said AMI has its roots as an internal FAIR project and has had support from company leaders, including Mark Zuckerberg and CTO Andrew Bosworth. But as Meta’s AI priorities shift toward catching up with competitors, the environment becomes less favorable.
At that point, it made sense “to leave the company, try to accelerate the development process, and build the things we can build with it,” LeCun said. “And I realized I could also make enough money to keep this going.”
In March, AMI Labs raised a $1.03 billion seed round, the largest in European startup history. Investors include Nvidia, Jeff Bezos, Eric Schmidt and others. With LeCun serving as CEO, the startup is led by CEO Alex LeBrun, a former engineering lead at FAIR and co-founder of health AI company Nabla, AMI Labs’ first partner. The company has assembled a heavy research team with experience at Meta, Google DeepMind, and other leading AI labs.
AMI says it aims to build applications in areas where this approach is important, including industrial process control, automation, wearables, robotics, and healthcare. “It won’t happen next year,” Lacon said. “We’re not going to have a country of data center geniuses next year, okay?”
At VivaTech, the discussion also expanded to include a larger question: Who should control the next generation of AI? If future assistants are built only by “a handful of companies” in Silicon Valley or China, “culture is in big trouble” and “democracy is in big trouble,” Lukon warned. Banning open source models in the name of safety ignores the role AI can play in spreading knowledge, he said.
This position also explains LeCun’s work as chief scientific advisor for the Tapestry Project, an effort led by the AI Alliance to build a shared global model for artificial intelligence through a distributed network of contributors. The AI Alliance is a non-profit group, jointly launched in 2023 by Meta and IBM, to support the development of open source AI. With Tapestry, the goal is to allow countries, companies, universities and others to contribute data or computing power without giving up control of that data. As AI begins to mediate what people read, watch, and search for, Lukon argues that people need alternatives to what some dominant American or Chinese company decides to build.
LeCun’s comments also come at a time when the Meta’s stance on open source has become more complicated. Llama 2, released in 2023 for free commercial use, makes Meta a rare counterweight for big tech companies to more closed-minded AI competitors. But Zuckerberg did since pointed out Meta may not release all future “super-intelligence” models publicly.
“We need access to a wide range of AI assistants for the same reason we need access to a wide range of journalism to get multiple sources of information,” LeCun said. “The only way I can see that this could happen is if there was an open, free foundation model, on top of which anyone could build their own specialized assistant for their own language or languages, their culture, their own value system, their political biases, and their centers of interest. And so there should be open source.”
