Yann LeCun, chief AI scientist at Meta and A Turing Award A Nobel Prize winner, he has long been a central figure in artificial intelligence. However, over the past year, his work has not only continued to push the boundaries of AI research, but has also sparked critical discussions about how society should address the opportunities and risks posed by this transformative technology.

Born in 1960 in Suezy-sous-Montmorency, France, LeCun has been a driving force in the field of artificial intelligence innovation. From being founding director of NYU's Center for Data Science in 2012, and co-founder of Meta AI in 2013, to shaping the future of open source AI, LeCun's practical vision makes him Emerge's Person of the Year.

"On the technical side, he was a visionary. There are only two people you can honestly say that about, and he's one of them," New York University computer science professor Rob Fergus He said Decryption In an interview. “More recently, his advocacy for open source and open research has been crucial to the Cambrian explosion of startups and people relying on these big language models.”

Fergus is an American computer scientist specializing in machine learning, deep learning, and generative models. He is a professor at New York University's Courant Institute and a Google DeepMind researcher, and co-founded Meta AI (formerly Facebook AI Research) with Yann LeCun in September 2013.

LeCun's influence on AI stretches back decades and includes his pioneering work in machine learning and neural networks. former Professor Silver At NYU, he has long advocated self-supervised learning, an approach inspired by how humans learn from their surroundings. In 2024, this vision has led to advances in artificial intelligence systems that can perceive, think, and plan with increasing sophistication, much like living organisms.

"There was a point around 2015 when reinforcement learning was seen as the path to artificial general intelligence. Yan had an analogy with the cake: unsupervised learning is the flesh, supervised learning is the frosting, and reinforcement learning is the cherry on top," he recalls. Professor Fergus. “This was scoffed at by many at the time, but it has been proven to be true. Modern LLM students are trained primarily through unsupervised learning, fine-tuned with minimal supervised data, and optimized using reinforcement learning based on human preferences.

Whether developing cutting-edge systems such as Meta's large open source language models, including Llama AI, or addressing the ethical and regulatory challenges of AI, LeCun has become a central figure in the global debate about the role of AI.

Professor Fergus said: "It was great to see him up close and all the wonderful things he did. More people should be listening to him."

Artificial intelligence regulations

One of LeCun's most controversial stances this year has been his outspoken opposition to to organize Basic AI models.

“He told me that he doesn't think AI regulations aren't necessary or the right thing,” the New York University mathematics professor said Russell Cavlish He said Decryption. “I think he's an optimist, and he sees all the good things that can come from artificial intelligence.”

Cavlish, director of the Courant Institute for Mathematical Sciences at New York University, has known Professor Licon since 2008, and has witnessed the rise of modern machine learning.

In June, LeCun took to X to assert that regulating models themselves could stifle innovation and hinder technological progress.

“Holding technology developers responsible for poor uses of products built on their technology will simply stop the development of the technology,” says LeCun. He said. “This will definitely stop the distribution of open source AI platforms, which will kill the entire AI ecosystem, not just startups, but also academic research.”

LeCun called for regulations to focus on applications where risks are more context-specific and manageable, and spoke out against regulating basic AI models, suggesting that regulating applications rather than the underlying technology would be more beneficial.

“Jan did the foundational work that made AI successful,” Cavlisch said. “His current relevance is that he is friendly, articulate and has a vision to advance AI towards artificial general intelligence.”

Criticism for fearmongering about artificial intelligence

LeCun has been outspoken in confronting what he sees as exaggerated fears surrounding the potential dangers of artificial intelligence.

“He doesn't give in to fearmongering, and he's optimistic about artificial intelligence, but he's not encouraging either,” Cavlish said. “He also promoted a path to improving this through robotics, by collecting data from the physical world.”

In an April appearance on Lex Friedman PodcastHe rejected the catastrophic predictions often associated with runaway superintelligence or uncontrolled artificial intelligence systems.

“AI imagines all kinds of disaster scenarios about how AI can escape or take control and kill us all, and that's based on a whole bunch of assumptions that are mostly false,” LeCun said. “The first assumption is that the emergence of superintelligence could be an event that we will encounter at some point, we will discover the secret, we will turn on a superintelligent machine, and because we haven't done it before, it will take over the world and kill us all. This is not true.

Since ChatGPT's launch in November 2022, the world has entered what many call an AI arms race. It was enabled by a century of Hollywood movies predicting the coming robot Resurrection We were helped by the news that AI developers are working with... United States Government And its allies to integrate artificial intelligence into their frameworks, many fear that artificial superintelligence will take over the world.

However, LeCun disagrees with these views, saying that the smartest AI would only have the level of intelligence of a small animal and not the global collective mind of the Matrix.

"It won't be an event. "We'll have intelligent systems like cats, all of which have all the characteristics of human intelligence, but their level of intelligence will be like a cat or a parrot," Lacon continued. "Then we'll work on making these things even smarter. As we make them smarter, we'll also put some guardrails in them.

In a hypothetical doomsday scenario where rogue AIs emerge, LeCun suggested that if developers cannot agree on how to control the AI ​​and one of them goes rogue, "good" AI could be deployed to fight the rogue systems.

The way forward for artificial intelligence

LeCun defends what he calls "Goal-oriented artificial intelligence,“AI systems don't just predict sequences or generate content, they are goal-driven and can understand, predict, and interact with the world as deeply as living organisms. This process involves creating AI systems that develop “global models”—internal representations of how things work. - Which enables causal thinking and the ability to plan and adapt strategies in real time.

LeCun has long been a proponent of self-supervised learning as a way to advance AI toward a more autonomous and general intelligence. He envisions AI learning how to perceive, think, and plan at multiple levels of abstraction, allowing it to learn from vast amounts of unlabeled data, similar to the way humans learn from their environment.

“The true AI revolution has not yet arrived,” LeCun said He said During a speech at the World Science and Technology Forum 2024 in Seoul. “In the near future, all of our interactions with the digital world will be mediated by AI assistants.”

Yann Licon's contributions to AI in 2024 reflect a drive for technological innovation and practical insight. His opposition to strict AI regulation, and his rejection of alarmist AI narratives, highlight his commitment to moving the field forward. As artificial intelligence continues to evolve, LeCun's influence ensures that it remains a force for technological advancement.