Yann LeCun

Yann LeCun invented the modern convolutional neural network at Bell Labs in the late 1980s, building systems that read 10% of all US checks by the late 1990s. He co-founded Meta's FAIR lab in 2013 and shared the 2018 Turing Award. He champions self-supervised learning, energy-based models, and his JEPA (Joint-Embedding Predictive Architecture) world-model program - arguing that autoregressive LLMs alone cannot reach human-level reasoning.

• Meta AI (FAIR)

  Chief AI Scientist · 2013–present

• NYU

  Silver Professor · 2003–present

• Convolutional Neural Networks (LeNet, 1989)

  Created the architecture underlying nearly all modern vision systems.

• MNIST benchmark

  Curated the most-used dataset in ML history.

• JEPA / World Models

  Proposed self-supervised joint-embedding predictive architectures as a path beyond LLMs.

• FAIR open research

  Established Meta's open-publishing AI lab; drove the release of LLaMA.

• Backpropagation Applied to Handwritten Zip Code Recognition

  Neural Computation · 1989

• Gradient-Based Learning Applied to Document Recognition

  Proceedings of the IEEE · 1998

• Deep Learning

  Nature · 2015

• A Path Towards Autonomous Machine Intelligence

  OpenReview · 2022

• Turing Award · 2018
• IEEE Neural Network Pioneer Award · 2014
• Legion of Honour · 2023

1. 1987

   PhD from Université Pierre et Marie Curie.

2. 1989

   Built LeNet, the first practical convolutional network, at Bell Labs.

3. 2013

   Founded Facebook AI Research (FAIR).

4. 2018

   Awarded the Turing Award.

5. 2022

   Published JEPA world-model manifesto.

6. 2024

   Public advocate that open foundation models reduce, not increase, risk.

• LLMs alone will not reach human-level intelligence.
• Open-source foundation models are the safer path.
• Existential AI risk is overstated relative to misuse risk.