Yann LeCun: Turing Award Winner and LLM Skeptic

👁 118 reads

Yann LeCun is one of the three “godfathers” of deep learning (2018 Turing Award alongside Hinton and Bengio), creator of the convolutional neural networks that read bank checks in the 1990s, Meta’s chief AI scientist for 12 years, and since December 2025 co-founder of his own company Advanced Machine Intelligence Labs. He is also the industry’s loudest skeptic of the thesis that large language models lead to human-level intelligence. Facts checked September 18, 2026.

Bell Labs: Networks That Read Checks

In his 1987 PhD thesis, LeCun proposed an early version of backpropagation, the algorithm that allows neural networks to learn, then spent a year working with Geoffrey Hinton in Toronto. In 1988 he joined Bell Labs in New Jersey, where he created LeNet, a convolutional network inspired by the structure of the visual cortex, deployed by NCR and banks to read handwritten checks. From 1996 he led image processing research at AT&T Labs and co-created the DjVu compression format, still used today by Internet Archive. In 2003 he moved to NYU.

Meta and the Turing Award

In 2013 Mark Zuckerberg hired him to build Facebook AI Research (FAIR), which for a decade was one of the three most important AI research centers alongside Google and OpenAI and released the open Llama models. In 2018 LeCun, Hinton and Yoshua Bengio received the Turing Award, computing’s Nobel Prize, for deep learning. In December 2025 he left Meta and co-founded Advanced Machine Intelligence Labs, a company building world models instead of language models.

Why LeCun Argues With the Rest of the Industry

LeCun has argued for years that large language models are a “dead end” on the path to human-level intelligence: they predict the next word but don’t understand the physical world, don’t plan, and have no persistent memory. His answer is energy-based models and the JEPA architecture, learning from video like a child observing its surroundings. He is also in sharp dispute with Hinton and Dario Amodei about AI existential risk, which he considers overblown, and supports open model weights. I map out who argues with whom in this industry on the Who’s Who in AI page.

On the other side of the debate: Yoshua Bengio, winner of the same Turing Award, assesses the risk very differently.

Frequently Asked Questions

How old is Yann LeCun?

66. He was born July 8, 1960 near Paris.

Does LeCun still work at Meta?

No. He left in December 2025 after twelve years and co-founded Advanced Machine Intelligence Labs. He remains a professor at NYU.

What did he win the Turing Award for?

For conceptual and engineering breakthroughs that made deep neural networks a critical component of computing, in particular for convolutional networks. He shared the 2018 award with Geoffrey Hinton and Yoshua Bengio.

Sources checked September 18, 2026: Wikipedia EN “Yann LeCun” with citations (ACM, Bell Labs, NYU, Meta and AMI Labs announcements).

// read moreDemis Hassabis: Google DeepMind CEO, Nobel Winner & AI PioneerMagazineGeoffrey Hinton: AI Godfather, Nobel Winner Who Left GoogleMagazineBest Loss Function for Tabular Regression: Study FindingsBenchmarks

Kinetyra