Meta AI veteran Yann LeCun raises $1 billion for AMI, betting on world models to rival today’s text-based AI systems.
Yann LeCun, formerly chief AI scientist at Meta, has raised more than $1 billion in seed money for his new startup, Advanced Machine Intelligence (AMI). His investors include European heavyweights (Hiro Capital, Cathay Innovation, Groupe Industriel Marcel Dassault), US behemoths (Nvidia, Bezos Expeditions, Eric Schmidt), and Asian giants (Toyota, Samsung).
The 65-year-old French researcher, who with two other professors won the 2018 Turing award—often called the Nobel Prize of computer science—spent 12 years leading Meta’s AI research. But when CEO Mark Zuckerberg decided that Alexandr Wang, 28, founder of Scale AI, would supervise the company’s AI strategy, LeCun knew that it was time to move on.
Headquartered in Paris, his new venture has global ambitions. The chatbots developed by OpenAI, Google, and Anthropic are, in LeCun’s eyes, limited to texts and do not grasp the real world.
AMI opts to focus on building “world models” that can better understand how the world works, achieve complex tasks, and anticipate outcomes. In three to five years, LeCun says he hopes to create apps for airplane engines, autos, and robots.
A professor of computer science at New York University, LeCun has experienced ups and downs in his long career, half of it spent in the US at ATT’s Bell Labs and Meta. In 1985, while studying at the Sorbonne, he published an early version of the learning algorithm, now known as error backpropagation, written in French. It flopped. US scientists attracted attention on the same subject one year later.
LeCun persisted, developing new machine learning methods at Bell Labs that enabled handwriting recognition of bank checks, deployed by NCR. He is also behind Djvu, an image compression technology. And he won the Turing award with two Canadian colleagues for their work on deep learning.
