This will also interest you
[EN VIDÉO] Interview: How was artificial intelligence born? Artificial intelligence aims to mimic the way the human brain works, or at least its logic…
ChatGPT, launched a year ago, does not impress him because he imagined it before anyone else: for thirty years, the Frenchman Yann LeCun has been betting everything on deep learning, deep learning that trains computers instead of programming them. His bet made the visionary a world star in artificial intelligence, the “father of AI”. But he is already thinking about the day after tomorrow, far beyond the popular text and image generators that are already outdated for him.
5 things you don’t know about deep learning
In 30 years, despite the doubts of many of his colleagues, this pioneer has ensured that his research area has established itself as the most innovative of the century and is at the heart of all AI today. Language models, content-generating AI, but also autonomous cars, facial recognition, predictive diagnosis… so many technologies that result from this.
With chatGPT and other generative AI launched in November 2022, the general public is also familiar with the concept and has recognized the potential of learning computers. As the highest recognition, Breton received the Turing Prize in 2019, the equivalent of the Nobel Prize for computer science.
After training at Pierre and Marie Curie University in Paris, then as a professor in New York, this cheerful, quick-thinking sexagenarian was personally recruited by Mark Zuckerberg in the hallways of a conference in 2013. Since then he has headed the AI research laboratory of the Meta Group (Facebook, Instagram).
On the way to becoming a “learning machine”
At the age of nine, he was influenced by Stanley Kubrick’s 2001 film “A Space Odyssey” about the dangers of a conscious computer, and he developed a passion for the first personal computers at a very early age. At the age of 21, the debate between the American linguist Noam Chomsky and the Swiss psychologist Jean Piaget about the nature and promotion of intelligence led him down the path of “learning machines”.
From artificial intelligence to collective intelligence
He was hired by Bell Labs in 1988 and attempted to apply this new technique to handwriting recognition on handwritten checks. Success rate: 50%. Bell throws in the towel. The technology is considered complex and too demanding in terms of computing power. Yann LeCun remains. In 2003, along with researchers Geoffrey Hinton and Yoshua Bengio, he fomented a so-called “neural network conspiracy” to revive interest in the scientific community through conferences. In 2012, their efforts finally paid off: Hinton and other scientists won an image recognition competition using a deep learning-based program.
The mass of data available on the Internet and the increasing power of microprocessors will make his dream possible. “Overnight, people stopped everything they were doing to use these models,” recalls Yann LeCun for Libération. “I’ve never seen that in science.”
An optimistic and enthusiastic pioneer who makes his difference
So much so that he scares some of his colleagues a little. In May 2023, his colleague and Turing Prize co-winner Geoffrey Hinton, 75, left Google, saying he “regretted” his invention, which “could pose a risk to humanity.” This summer, a collective of scientists, bosses and experts – including big names Elon Musk and Sam Altman – called for a six-month pause in AI research, which they say threatens humanity’s existence.
This does not apply to Yann LeCun, a rationalist atheist who remains convinced of the benefits of AI and progress in general. “The very idea of wanting to slow down AI research is like a new obscurantism,” he said. According to him, technology leaders want to “sow fear” to maintain the monopoly and their business.
“People have common sense, but machines don’t.”
Bucking the global trend, he emphasizes his reservations about generative AI. “Today’s AI and machine learning really suck. Humans have common sense, but machines do not. I am not afraid that AI will escape our control and lead to the destruction of humanity.”
ChatGPT? For him it is “a dead end”, a statistical prediction model that will no longer be used “in 5 years”. However, his employer Meta brought similar AIs to market. But Yann LeCun maintains his distinction. And he believes in it as hard as he can. “With the help of AI, the intelligence and creativity of every individual is strengthened. This can lead to a new age of enlightenment.”