👁️Yann LeCun
Original nameYann LeCun
French-American Computer Scientist, Founder of the Convolutional Neural Network
The Laws Behind the Technology · Theoretical Founders
The Convolutional Neural Network (CNN)"LeNet" Handwritten-Digit RecognitionThe 2018 Turing Award
Who they are
Yann LeCun (1960– ) is a French-American computer scientist who in 1989, while working at Bell Labs, combined the backpropagation algorithm with a network structure imitating the local-perception feature of the biological visual cortex, designing the "convolutional neural network" (CNN) and training a handwritten-digit recognition system called "LeNet," a system thereafter actually deployed by several US banks to automatically recognize handwritten amounts and postal codes on checks, one of the earliest cases of large-scale commercial application of deep neural networks. The convolutional neural network, through designs such as local receptive fields and weight sharing, greatly reduced the number of parameters needed to process high-dimensional data like images, and this architecture thereafter became the structural basis of "AlexNet" in 2012 and almost all later computer-vision deep-learning models. LeCun thereafter long served as a professor at New York University and from 2013 concurrently as chief AI scientist at Facebook (later renamed Meta), leading the company’s foundational AI research. For these foundational contributions he, with Hinton and Bengio, won the 2018 Turing Award, the three thereby called the "three giants of deep learning." Unlike Hinton and Bengio, who in recent years have taken a more cautious attitude toward the potential risks of AI, LeCun’s public stance is relatively optimistic, inclining to stress the limits of the current large-language-model technical path, and this divergence of stance is also one of the topics of public discussion among the three Turing laureates in recent years.
Primary sourcesLeCun et al., "Gradient-Based Learning Applied to Document Recognition" (1998, the LeNet-5 paper)
Key stories
Letting the Machine Read Handwritten Digits on Checks
The LeNet convolutional neural network LeCun designed, by imitating the human visual cortex’s processing of "local perception, layer-by-layer abstraction," could reliably recognize handwritten digits and letters, and this system was thereafter actually deployed by several US banks in check-processing systems to recognize the handwritten amounts on checks — an early case of deep neural network technology quietly entering everyday financial infrastructure and creating real commercial value before it was widely recognized.
The Line Divergence Within the "Three Giants"
LeCun, named with Hinton and Bengio the "three giants of deep learning," jointly won the 2018 Turing Award, but the three have in recent years publicly expressed not-wholly-consistent views on whether the current large-language-model technical path can lead to true general artificial intelligence and on the degree of risk this technology may bring, and this line divergence among the very scholars who together laid the foundation of deep learning is itself an interesting facet for observing the internal debate of the AI field.
Echoes today
Below are how modern works borrow or reinterpret this name or story — not the original material. The two differ, so keep them apart.
The public line divergence of the "three giants"LeCun has in recent years publicly expressed on social media many times a skeptical attitude toward whether "the current large-language-model path can lead to general AI," a certain contrast with the more risk-warning stance of Hinton and Bengio in recent years, and such public debates are an important window for the public to understand the internal technical-line dispute of the AI field.