Yann LeCun

Yann LeCun has spent four decades proving that the dominant AI paradigm of any given moment is rarely the final word — and in 2026 he is once again positioning himself against the mainstream, this time against the large language models that made his former employer, Meta, one of the industry's biggest spenders. Born near Paris in 1960, LeCun trained as an engineer at ESIEE Paris before completing a PhD in computer science, during which he developed early versions of backpropagation-based learning together with a postdoctoral stint under Geoffrey Hinton in Toronto.

At Bell Labs through the late 1980s and 1990s, LeCun built LeNet, the convolutional neural network architecture that could read handwritten digits well enough to be deployed reading personal cheques across the US banking system — one of AI's first genuine commercial deployments and the direct ancestor of the CNN architectures that underpin modern computer vision. A subsequent stint at AT&T Labs and a move to New York University, where he founded the NYU Center for Data Science in 2012, cemented his reputation as one of deep learning's three foundational architects.

In 2013, LeCun became founding director of Facebook AI Research and, later, Chief AI Scientist at Meta, where for over a decade he oversaw one of the world's largest corporate AI research operations while remaining an outspoken sceptic of the idea that scaling large language models alone would lead to human-level intelligence. He argued instead for 'world models' — systems that learn by predicting and simulating physical reality, closer to how animals and humans learn — a thesis that put him increasingly at odds with Meta's heavy generative-AI product push.

That disagreement came to a head in late 2025, when LeCun announced his departure from Meta to found Advanced Machine Intelligence Labs (AMI Labs), which he now leads as executive chair. The startup closed one of the largest seed rounds in history, reportedly over $1 billion, to pursue his world-model approach as an explicit alternative to the LLM-centred strategies of OpenAI, Anthropic, Google and his former employer, while he retains his post as a professor at NYU's Courant Institute.

LeCun shared the 2018 Turing Award with Geoffrey Hinton and Yoshua Bengio for their foundational contributions to deep learning, making all three inaugural giants of the field colleagues, occasional intellectual rivals and now, in different ways, critics of the industry's current direction. His willingness to bet his own venture on a contrarian scientific position — after a career spent inside the largest AI labs in the world — keeps him among the most closely watched figures in data science.